vscode69 commited on
Commit
99afdfe
·
1 Parent(s): 2fbbe44

second half

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .idea/TalkSHOW.iml +8 -0
  2. .idea/inspectionProfiles/profiles_settings.xml +6 -0
  3. .idea/modules.xml +8 -0
  4. .idea/vcs.xml +6 -0
  5. .idea/workspace.xml +55 -0
  6. config/LS3DCG.json +64 -0
  7. config/body_pixel.json +63 -0
  8. config/body_vq.json +62 -0
  9. config/face.json +59 -0
  10. data_utils/__init__.py +3 -0
  11. data_utils/__pycache__/__init__.cpython-37.pyc +0 -0
  12. data_utils/__pycache__/consts.cpython-37.pyc +0 -0
  13. data_utils/__pycache__/dataloader_torch.cpython-37.pyc +0 -0
  14. data_utils/__pycache__/lower_body.cpython-37.pyc +0 -0
  15. data_utils/__pycache__/mesh_dataset.cpython-37.pyc +0 -0
  16. data_utils/__pycache__/rotation_conversion.cpython-37.pyc +0 -0
  17. data_utils/__pycache__/utils.cpython-37.pyc +0 -0
  18. data_utils/apply_split.py +51 -0
  19. data_utils/axis2matrix.py +29 -0
  20. data_utils/consts.py +0 -0
  21. data_utils/dataloader_torch.py +279 -0
  22. data_utils/dataset_preprocess.py +170 -0
  23. data_utils/get_j.py +51 -0
  24. data_utils/hand_component.json +0 -0
  25. data_utils/lower_body.py +143 -0
  26. data_utils/mesh_dataset.py +348 -0
  27. data_utils/rotation_conversion.py +551 -0
  28. data_utils/split_more_than_2s.pkl +3 -0
  29. data_utils/split_train_val_test.py +27 -0
  30. data_utils/train_val_test.json +0 -0
  31. data_utils/utils.py +318 -0
  32. evaluation/FGD.py +199 -0
  33. evaluation/__init__.py +0 -0
  34. evaluation/__pycache__/__init__.cpython-37.pyc +0 -0
  35. evaluation/__pycache__/metrics.cpython-37.pyc +0 -0
  36. evaluation/diversity_LVD.py +64 -0
  37. evaluation/get_quality_samples.py +62 -0
  38. evaluation/metrics.py +109 -0
  39. evaluation/mode_transition.py +60 -0
  40. evaluation/peak_velocity.py +65 -0
  41. evaluation/util.py +148 -0
  42. losses/__init__.py +1 -0
  43. losses/__pycache__/__init__.cpython-37.pyc +0 -0
  44. losses/__pycache__/losses.cpython-37.pyc +0 -0
  45. losses/losses.py +91 -0
  46. nets/LS3DCG.py +414 -0
  47. nets/__init__.py +8 -0
  48. nets/__pycache__/__init__.cpython-37.pyc +0 -0
  49. nets/__pycache__/base.cpython-37.pyc +0 -0
  50. nets/__pycache__/init_model.cpython-37.pyc +0 -0
.idea/TalkSHOW.iml ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <module type="PYTHON_MODULE" version="4">
3
+ <component name="NewModuleRootManager">
4
+ <content url="file://$MODULE_DIR$" />
5
+ <orderEntry type="inheritedJdk" />
6
+ <orderEntry type="sourceFolder" forTests="false" />
7
+ </component>
8
+ </module>
.idea/inspectionProfiles/profiles_settings.xml ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ <component name="InspectionProjectProfileManager">
2
+ <settings>
3
+ <option name="USE_PROJECT_PROFILE" value="false" />
4
+ <version value="1.0" />
5
+ </settings>
6
+ </component>
.idea/modules.xml ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <project version="4">
3
+ <component name="ProjectModuleManager">
4
+ <modules>
5
+ <module fileurl="file://$PROJECT_DIR$/.idea/TalkSHOW.iml" filepath="$PROJECT_DIR$/.idea/TalkSHOW.iml" />
6
+ </modules>
7
+ </component>
8
+ </project>
.idea/vcs.xml ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <project version="4">
3
+ <component name="VcsDirectoryMappings">
4
+ <mapping directory="" vcs="Git" />
5
+ </component>
6
+ </project>
.idea/workspace.xml ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <project version="4">
3
+ <component name="ChangeListManager">
4
+ <list default="true" id="210ec380-7aeb-4dbc-977d-b398d2f30cdf" name="Changes" comment="" />
5
+ <option name="SHOW_DIALOG" value="false" />
6
+ <option name="HIGHLIGHT_CONFLICTS" value="true" />
7
+ <option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
8
+ <option name="LAST_RESOLUTION" value="IGNORE" />
9
+ </component>
10
+ <component name="Git.Settings">
11
+ <option name="RECENT_GIT_ROOT_PATH" value="$PROJECT_DIR$" />
12
+ </component>
13
+ <component name="GitHubPullRequestSearchHistory"><![CDATA[{
14
+ "lastFilter": {
15
+ "state": "OPEN",
16
+ "assignee": "khaitha"
17
+ }
18
+ }]]></component>
19
+ <component name="ProjectColorInfo"><![CDATA[{
20
+ "associatedIndex": 5
21
+ }]]></component>
22
+ <component name="ProjectId" id="2nX5jW7aFzy8GxYLTCwAZXC0ETa" />
23
+ <component name="ProjectViewState">
24
+ <option name="hideEmptyMiddlePackages" value="true" />
25
+ <option name="showLibraryContents" value="true" />
26
+ </component>
27
+ <component name="PropertiesComponent"><![CDATA[{
28
+ "keyToString": {
29
+ "RunOnceActivity.ShowReadmeOnStart": "true",
30
+ "git-widget-placeholder": "main",
31
+ "nodejs_package_manager_path": "npm",
32
+ "vue.rearranger.settings.migration": "true"
33
+ }
34
+ }]]></component>
35
+ <component name="SharedIndexes">
36
+ <attachedChunks>
37
+ <set>
38
+ <option value="bundled-js-predefined-d6986cc7102b-5c90d61e3bab-JavaScript-PY-242.23339.19" />
39
+ <option value="bundled-python-sdk-0029f7779945-399fe30bd8c1-com.jetbrains.pycharm.pro.sharedIndexes.bundled-PY-242.23339.19" />
40
+ </set>
41
+ </attachedChunks>
42
+ </component>
43
+ <component name="SpellCheckerSettings" RuntimeDictionaries="0" Folders="0" CustomDictionaries="0" DefaultDictionary="application-level" UseSingleDictionary="true" transferred="true" />
44
+ <component name="TaskManager">
45
+ <task active="true" id="Default" summary="Default task">
46
+ <changelist id="210ec380-7aeb-4dbc-977d-b398d2f30cdf" name="Changes" comment="" />
47
+ <created>1729106717261</created>
48
+ <option name="number" value="Default" />
49
+ <option name="presentableId" value="Default" />
50
+ <updated>1729106717261</updated>
51
+ <workItem from="1729106719428" duration="5000" />
52
+ </task>
53
+ <servers />
54
+ </component>
55
+ </project>
config/LS3DCG.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config_root_path": "/is/cluster/scratch/hyi/ExpressiveBody/SMPLifyX4/scripts",
3
+ "dataset_load_mode": "pickle",
4
+ "store_file_path": "store.pkl",
5
+ "smplx_npz_path": "visualise/smplx_model/SMPLX_NEUTRAL_2020.npz",
6
+ "extra_joint_path": "visualise/smplx_model/smplx_extra_joints.yaml",
7
+ "j14_regressor_path": "visualise/smplx_model/SMPLX_to_J14.pkl",
8
+ "param": {
9
+ "w_j": 1,
10
+ "w_b": 1,
11
+ "w_h": 1
12
+ },
13
+ "Data": {
14
+ "data_root": "../ExpressiveWholeBodyDatasetv1.0/",
15
+ "pklname": "_3d_mfcc.pkl",
16
+ "whole_video": false,
17
+ "pose": {
18
+ "normalization": false,
19
+ "convert_to_6d": false,
20
+ "norm_method": "all",
21
+ "augmentation": false,
22
+ "generate_length": 88,
23
+ "pre_pose_length": 0,
24
+ "pose_dim": 99,
25
+ "expression": true
26
+ },
27
+ "aud": {
28
+ "feat_method": "mfcc",
29
+ "aud_feat_dim": 64,
30
+ "aud_feat_win_size": null,
31
+ "context_info": false
32
+ }
33
+ },
34
+ "Model": {
35
+ "model_type": "body",
36
+ "model_name": "s2g_LS3DCG",
37
+ "code_num": 2048,
38
+ "AudioOpt": "Adam",
39
+ "encoder_choice": "mfcc",
40
+ "gan": false
41
+ },
42
+ "DataLoader": {
43
+ "batch_size": 128,
44
+ "num_workers": 0
45
+ },
46
+ "Train": {
47
+ "epochs": 100,
48
+ "max_gradient_norm": 5,
49
+ "learning_rate": {
50
+ "generator_learning_rate": 1e-4,
51
+ "discriminator_learning_rate": 1e-4
52
+ },
53
+ "weights": {
54
+ "keypoint_loss_weight": 1.0,
55
+ "gan_loss_weight": 1.0
56
+ }
57
+ },
58
+ "Log": {
59
+ "save_every": 50,
60
+ "print_every": 200,
61
+ "name": "LS3DCG"
62
+ }
63
+ }
64
+
config/body_pixel.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config_root_path": "/is/cluster/scratch/hyi/ExpressiveBody/SMPLifyX4/scripts",
3
+ "dataset_load_mode": "json",
4
+ "store_file_path": "store.pkl",
5
+ "smplx_npz_path": "visualise/smplx_model/SMPLX_NEUTRAL_2020.npz",
6
+ "extra_joint_path": "visualise/smplx_model/smplx_extra_joints.yaml",
7
+ "j14_regressor_path": "visualise/smplx_model/SMPLX_to_J14.pkl",
8
+ "param": {
9
+ "w_j": 1,
10
+ "w_b": 1,
11
+ "w_h": 1
12
+ },
13
+ "Data": {
14
+ "data_root": "../ExpressiveWholeBodyDatasetv1.0/",
15
+ "pklname": "_3d_mfcc.pkl",
16
+ "whole_video": false,
17
+ "pose": {
18
+ "normalization": false,
19
+ "convert_to_6d": false,
20
+ "norm_method": "all",
21
+ "augmentation": false,
22
+ "generate_length": 88,
23
+ "pre_pose_length": 0,
24
+ "pose_dim": 99,
25
+ "expression": true
26
+ },
27
+ "aud": {
28
+ "feat_method": "mfcc",
29
+ "aud_feat_dim": 64,
30
+ "aud_feat_win_size": null,
31
+ "context_info": false
32
+ }
33
+ },
34
+ "Model": {
35
+ "model_type": "body",
36
+ "model_name": "s2g_body_pixel",
37
+ "composition": true,
38
+ "code_num": 2048,
39
+ "bh_model": true,
40
+ "AudioOpt": "Adam",
41
+ "encoder_choice": "mfcc",
42
+ "gan": false,
43
+ "vq_path": "./experiments/2022-10-31-smplx_S2G-body-vq-3d/ckpt-99.pth"
44
+ },
45
+ "DataLoader": {
46
+ "batch_size": 128,
47
+ "num_workers": 0
48
+ },
49
+ "Train": {
50
+ "epochs": 100,
51
+ "max_gradient_norm": 5,
52
+ "learning_rate": {
53
+ "generator_learning_rate": 1e-4,
54
+ "discriminator_learning_rate": 1e-4
55
+ }
56
+ },
57
+ "Log": {
58
+ "save_every": 50,
59
+ "print_every": 200,
60
+ "name": "body-pixel2"
61
+ }
62
+ }
63
+
config/body_vq.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config_root_path": "/is/cluster/scratch/hyi/ExpressiveBody/SMPLifyX4/scripts",
3
+ "dataset_load_mode": "json",
4
+ "store_file_path": "store.pkl",
5
+ "smplx_npz_path": "visualise/smplx_model/SMPLX_NEUTRAL_2020.npz",
6
+ "extra_joint_path": "visualise/smplx_model/smplx_extra_joints.yaml",
7
+ "j14_regressor_path": "visualise/smplx_model/SMPLX_to_J14.pkl",
8
+ "param": {
9
+ "w_j": 1,
10
+ "w_b": 1,
11
+ "w_h": 1
12
+ },
13
+ "Data": {
14
+ "data_root": "../ExpressiveWholeBodyDatasetv1.0/",
15
+ "pklname": "_3d_mfcc.pkl",
16
+ "whole_video": false,
17
+ "pose": {
18
+ "normalization": false,
19
+ "convert_to_6d": false,
20
+ "norm_method": "all",
21
+ "augmentation": false,
22
+ "generate_length": 88,
23
+ "pre_pose_length": 0,
24
+ "pose_dim": 99,
25
+ "expression": true
26
+ },
27
+ "aud": {
28
+ "feat_method": "mfcc",
29
+ "aud_feat_dim": 64,
30
+ "aud_feat_win_size": null,
31
+ "context_info": false
32
+ }
33
+ },
34
+ "Model": {
35
+ "model_type": "body",
36
+ "model_name": "s2g_body_vq",
37
+ "composition": true,
38
+ "code_num": 2048,
39
+ "bh_model": true,
40
+ "AudioOpt": "Adam",
41
+ "encoder_choice": "mfcc",
42
+ "gan": false
43
+ },
44
+ "DataLoader": {
45
+ "batch_size": 128,
46
+ "num_workers": 0
47
+ },
48
+ "Train": {
49
+ "epochs": 100,
50
+ "max_gradient_norm": 5,
51
+ "learning_rate": {
52
+ "generator_learning_rate": 1e-4,
53
+ "discriminator_learning_rate": 1e-4
54
+ }
55
+ },
56
+ "Log": {
57
+ "save_every": 50,
58
+ "print_every": 200,
59
+ "name": "body-vq"
60
+ }
61
+ }
62
+
config/face.json ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config_root_path": "/is/cluster/scratch/hyi/ExpressiveBody/SMPLifyX4/scripts",
3
+ "dataset_load_mode": "json",
4
+ "store_file_path": "store.pkl",
5
+ "smplx_npz_path": "visualise/smplx_model/SMPLX_NEUTRAL_2020.npz",
6
+ "extra_joint_path": "visualise/smplx_model/smplx_extra_joints.yaml",
7
+ "j14_regressor_path": "visualise/smplx_model/SMPLX_to_J14.pkl",
8
+ "param": {
9
+ "w_j": 1,
10
+ "w_b": 1,
11
+ "w_h": 1
12
+ },
13
+ "Data": {
14
+ "data_root": "../ExpressiveWholeBodyDatasetv1.0/",
15
+ "pklname": "_3d_wv2.pkl",
16
+ "whole_video": true,
17
+ "pose": {
18
+ "normalization": false,
19
+ "convert_to_6d": false,
20
+ "norm_method": "all",
21
+ "augmentation": false,
22
+ "generate_length": 88,
23
+ "pre_pose_length": 0,
24
+ "pose_dim": 99,
25
+ "expression": true
26
+ },
27
+ "aud": {
28
+ "feat_method": "mfcc",
29
+ "aud_feat_dim": 64,
30
+ "aud_feat_win_size": null,
31
+ "context_info": false
32
+ }
33
+ },
34
+ "Model": {
35
+ "model_type": "face",
36
+ "model_name": "s2g_face",
37
+ "AudioOpt": "SGD",
38
+ "encoder_choice": "faceformer",
39
+ "gan": false
40
+ },
41
+ "DataLoader": {
42
+ "batch_size": 1,
43
+ "num_workers": 0
44
+ },
45
+ "Train": {
46
+ "epochs": 100,
47
+ "max_gradient_norm": 5,
48
+ "learning_rate": {
49
+ "generator_learning_rate": 1e-4,
50
+ "discriminator_learning_rate": 1e-4
51
+ }
52
+ },
53
+ "Log": {
54
+ "save_every": 50,
55
+ "print_every": 1000,
56
+ "name": "face"
57
+ }
58
+ }
59
+
data_utils/__init__.py ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ # from .dataloader_csv import MultiVidData as csv_data
2
+ from .dataloader_torch import MultiVidData as torch_data
3
+ from .utils import get_melspec, get_mfcc, get_mfcc_old, get_mfcc_psf, get_mfcc_psf_min, get_mfcc_ta
data_utils/__pycache__/__init__.cpython-37.pyc ADDED
Binary file (375 Bytes). View file
 
data_utils/__pycache__/consts.cpython-37.pyc ADDED
Binary file (92.7 kB). View file
 
data_utils/__pycache__/dataloader_torch.cpython-37.pyc ADDED
Binary file (5.31 kB). View file
 
data_utils/__pycache__/lower_body.cpython-37.pyc ADDED
Binary file (3.91 kB). View file
 
data_utils/__pycache__/mesh_dataset.cpython-37.pyc ADDED
Binary file (7.9 kB). View file
 
data_utils/__pycache__/rotation_conversion.cpython-37.pyc ADDED
Binary file (16.4 kB). View file
 
data_utils/__pycache__/utils.cpython-37.pyc ADDED
Binary file (7.42 kB). View file
 
data_utils/apply_split.py ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ from tqdm import tqdm
3
+ import pickle
4
+ import shutil
5
+
6
+ speakers = ['seth', 'oliver', 'conan', 'chemistry']
7
+ source_data_root = "../expressive_body-V0.7"
8
+ data_root = "D:/Downloads/SHOW_dataset_v1.0/ExpressiveWholeBodyDatasetReleaseV1.0"
9
+
10
+ f_read = open('split_more_than_2s.pkl', 'rb')
11
+ f_save = open('none.pkl', 'wb')
12
+ data_split = pickle.load(f_read)
13
+ none_split = []
14
+
15
+ train = val = test = 0
16
+
17
+ for speaker_name in speakers:
18
+ speaker_root = os.path.join(data_root, speaker_name)
19
+
20
+ videos = [v for v in data_split[speaker_name]]
21
+
22
+ for vid in tqdm(videos, desc="Processing training data of {}......".format(speaker_name)):
23
+ for split in data_split[speaker_name][vid]:
24
+ for seq in data_split[speaker_name][vid][split]:
25
+
26
+ seq = seq.replace('\\', '/')
27
+ old_file_path = os.path.join(data_root, speaker_name, vid, seq.split('/')[-1])
28
+ old_file_path = old_file_path.replace('\\', '/')
29
+ new_file_path = seq.replace(source_data_root.split('/')[-1], data_root.split('/')[-1])
30
+ try:
31
+ shutil.move(old_file_path, new_file_path)
32
+ if split == 'train':
33
+ train = train + 1
34
+ elif split == 'test':
35
+ test = test + 1
36
+ elif split == 'val':
37
+ val = val + 1
38
+ except FileNotFoundError:
39
+ none_split.append(old_file_path)
40
+ print(f"The file {old_file_path} does not exists.")
41
+ except shutil.Error:
42
+ none_split.append(old_file_path)
43
+ print(f"The file {old_file_path} does not exists.")
44
+
45
+ print(none_split.__len__())
46
+ pickle.dump(none_split, f_save)
47
+ f_save.close()
48
+
49
+ print(train, val, test)
50
+
51
+
data_utils/axis2matrix.py ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import numpy as np
2
+ import math
3
+ import scipy.linalg as linalg
4
+
5
+
6
+ def rotate_mat(axis, radian):
7
+
8
+ a = np.cross(np.eye(3), axis / linalg.norm(axis) * radian)
9
+
10
+ rot_matrix = linalg.expm(a)
11
+
12
+ return rot_matrix
13
+
14
+ def aaa2mat(axis, sin, cos):
15
+ i = np.eye(3)
16
+ nnt = np.dot(axis.T, axis)
17
+ s = np.asarray([[0, -axis[0,2], axis[0,1]],
18
+ [axis[0,2], 0, -axis[0,0]],
19
+ [-axis[0,1], axis[0,0], 0]])
20
+ r = cos * i + (1-cos)*nnt +sin * s
21
+ return r
22
+
23
+ rand_axis = np.asarray([[1,0,0]])
24
+ #旋转角度
25
+ r = math.pi/2
26
+ #返回旋转矩阵
27
+ rot_matrix = rotate_mat(rand_axis, r)
28
+ r2 = aaa2mat(rand_axis, np.sin(r), np.cos(r))
29
+ print(rot_matrix)
data_utils/consts.py ADDED
The diff for this file is too large to render. See raw diff
 
data_utils/dataloader_torch.py ADDED
@@ -0,0 +1,279 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import sys
2
+ import os
3
+ sys.path.append(os.getcwd())
4
+ import os
5
+ from tqdm import tqdm
6
+ from data_utils.utils import *
7
+ import torch.utils.data as data
8
+ from data_utils.mesh_dataset import SmplxDataset
9
+ from transformers import Wav2Vec2Processor
10
+
11
+
12
+ class MultiVidData():
13
+ def __init__(self,
14
+ data_root,
15
+ speakers,
16
+ split='train',
17
+ limbscaling=False,
18
+ normalization=False,
19
+ norm_method='new',
20
+ split_trans_zero=False,
21
+ num_frames=25,
22
+ num_pre_frames=25,
23
+ num_generate_length=None,
24
+ aud_feat_win_size=None,
25
+ aud_feat_dim=64,
26
+ feat_method='mel_spec',
27
+ context_info=False,
28
+ smplx=False,
29
+ audio_sr=16000,
30
+ convert_to_6d=False,
31
+ expression=False,
32
+ config=None
33
+ ):
34
+ self.data_root = data_root
35
+ self.speakers = speakers
36
+ self.split = split
37
+ if split == 'pre':
38
+ self.split = 'train'
39
+ self.norm_method=norm_method
40
+ self.normalization = normalization
41
+ self.limbscaling = limbscaling
42
+ self.convert_to_6d = convert_to_6d
43
+ self.num_frames=num_frames
44
+ self.num_pre_frames=num_pre_frames
45
+ if num_generate_length is None:
46
+ self.num_generate_length = num_frames
47
+ else:
48
+ self.num_generate_length = num_generate_length
49
+ self.split_trans_zero=split_trans_zero
50
+
51
+ dataset = SmplxDataset
52
+
53
+ if self.split_trans_zero:
54
+ self.trans_dataset_list = []
55
+ self.zero_dataset_list = []
56
+ else:
57
+ self.all_dataset_list = []
58
+ self.dataset={}
59
+ self.complete_data=[]
60
+ self.config=config
61
+ load_mode=self.config.dataset_load_mode
62
+
63
+ ######################load with pickle file
64
+ if load_mode=='pickle':
65
+ import pickle
66
+ import subprocess
67
+
68
+ # store_file_path='/tmp/store.pkl'
69
+ # cp /is/cluster/scratch/hyi/ExpressiveBody/SMPLifyX4/scripts/store.pkl /tmp/store.pkl
70
+ # subprocess.run(f'cp /is/cluster/scratch/hyi/ExpressiveBody/SMPLifyX4/scripts/store.pkl {store_file_path}',shell=True)
71
+
72
+ # f = open(self.config.store_file_path, 'rb+')
73
+ f = open(self.split+config.Data.pklname, 'rb+')
74
+ self.dataset=pickle.load(f)
75
+ f.close()
76
+ for key in self.dataset:
77
+ self.complete_data.append(self.dataset[key].complete_data)
78
+ ######################load with pickle file
79
+
80
+ ######################load with a csv file
81
+ elif load_mode=='csv':
82
+
83
+ # 这里从我的一个code文件夹导入的,后续再完善进来
84
+ try:
85
+ sys.path.append(self.config.config_root_path)
86
+ from config import config_path
87
+ from csv_parser import csv_parse
88
+
89
+ except ImportError as e:
90
+ print(f'err: {e}')
91
+ raise ImportError('config root path error...')
92
+
93
+
94
+ for speaker_name in self.speakers:
95
+ # df_intervals=pd.read_csv(self.config.voca_csv_file_path)
96
+ df_intervals=None
97
+ df_intervals=df_intervals[df_intervals['speaker']==speaker_name]
98
+ df_intervals = df_intervals[df_intervals['dataset'] == self.split]
99
+
100
+ print(f'speaker {speaker_name} train interval length: {len(df_intervals)}')
101
+ for iter_index, (_, interval) in tqdm(
102
+ (enumerate(df_intervals.iterrows())),desc=f'load {speaker_name}'
103
+ ):
104
+
105
+ (
106
+ interval_index,
107
+ interval_speaker,
108
+ interval_video_fn,
109
+ interval_id,
110
+
111
+ start_time,
112
+ end_time,
113
+ duration_time,
114
+ start_time_10,
115
+ over_flow_flag,
116
+ short_dur_flag,
117
+
118
+ big_video_dir,
119
+ small_video_dir_name,
120
+ speaker_video_path,
121
+
122
+ voca_basename,
123
+ json_basename,
124
+ wav_basename,
125
+ voca_top_clip_path,
126
+ voca_json_clip_path,
127
+ voca_wav_clip_path,
128
+
129
+ audio_output_fn,
130
+ image_output_path,
131
+ pifpaf_output_path,
132
+ mp_output_path,
133
+ op_output_path,
134
+ deca_output_path,
135
+ pixie_output_path,
136
+ cam_output_path,
137
+ ours_output_path,
138
+ merge_output_path,
139
+ multi_output_path,
140
+ gt_output_path,
141
+ ours_images_path,
142
+ pkl_fil_path,
143
+ )=csv_parse(interval)
144
+
145
+ if not os.path.exists(pkl_fil_path) or not os.path.exists(audio_output_fn):
146
+ continue
147
+
148
+ key=f'{interval_video_fn}/{small_video_dir_name}'
149
+ self.dataset[key] = dataset(
150
+ data_root=pkl_fil_path,
151
+ speaker=speaker_name,
152
+ audio_fn=audio_output_fn,
153
+ audio_sr=audio_sr,
154
+ fps=num_frames,
155
+ feat_method=feat_method,
156
+ audio_feat_dim=aud_feat_dim,
157
+ train=(self.split == 'train'),
158
+ load_all=True,
159
+ split_trans_zero=self.split_trans_zero,
160
+ limbscaling=self.limbscaling,
161
+ num_frames=self.num_frames,
162
+ num_pre_frames=self.num_pre_frames,
163
+ num_generate_length=self.num_generate_length,
164
+ audio_feat_win_size=aud_feat_win_size,
165
+ context_info=context_info,
166
+ convert_to_6d=convert_to_6d,
167
+ expression=expression,
168
+ config=self.config
169
+ )
170
+ self.complete_data.append(self.dataset[key].complete_data)
171
+ ######################load with a csv file
172
+
173
+ ######################origin load method
174
+ elif load_mode=='json':
175
+
176
+ # if self.split == 'train':
177
+ # import pickle
178
+ # f = open('store.pkl', 'rb+')
179
+ # self.dataset=pickle.load(f)
180
+ # f.close()
181
+ # for key in self.dataset:
182
+ # self.complete_data.append(self.dataset[key].complete_data)
183
+ # else:https://pytorch-tutorial-assets.s3.amazonaws.com/VOiCES_devkit/source-16k/train/sp0307/Lab41-SRI-VOiCES-src-sp0307-ch127535-sg0042.wav
184
+ # if config.Model.model_type == 'face':
185
+ am = Wav2Vec2Processor.from_pretrained("vitouphy/wav2vec2-xls-r-300m-phoneme")
186
+ am_sr = 16000
187
+ # else:
188
+ # am, am_sr = None, None
189
+ for speaker_name in self.speakers:
190
+ speaker_root = os.path.join(self.data_root, speaker_name)
191
+
192
+ videos=[v for v in os.listdir(speaker_root) ]
193
+ print(videos)
194
+
195
+ haode = huaide = 0
196
+
197
+ for vid in tqdm(videos, desc="Processing training data of {}......".format(speaker_name)):
198
+ source_vid=vid
199
+ # vid_pth=os.path.join(speaker_root, source_vid, 'images/half', self.split)
200
+ vid_pth = os.path.join(speaker_root, source_vid, self.split)
201
+ if smplx == 'pose':
202
+ seqs = [s for s in os.listdir(vid_pth) if (s.startswith('clip'))]
203
+ else:
204
+ try:
205
+ seqs = [s for s in os.listdir(vid_pth)]
206
+ except:
207
+ continue
208
+
209
+ for s in seqs:
210
+ seq_root=os.path.join(vid_pth, s)
211
+ key = seq_root # correspond to clip******
212
+ audio_fname = os.path.join(speaker_root, source_vid, self.split, s, '%s.wav' % (s))
213
+ motion_fname = os.path.join(speaker_root, source_vid, self.split, s, '%s.pkl' % (s))
214
+ if not os.path.isfile(audio_fname) or not os.path.isfile(motion_fname):
215
+ huaide = huaide + 1
216
+ continue
217
+
218
+ self.dataset[key]=dataset(
219
+ data_root=seq_root,
220
+ speaker=speaker_name,
221
+ motion_fn=motion_fname,
222
+ audio_fn=audio_fname,
223
+ audio_sr=audio_sr,
224
+ fps=num_frames,
225
+ feat_method=feat_method,
226
+ audio_feat_dim=aud_feat_dim,
227
+ train=(self.split=='train'),
228
+ load_all=True,
229
+ split_trans_zero=self.split_trans_zero,
230
+ limbscaling=self.limbscaling,
231
+ num_frames=self.num_frames,
232
+ num_pre_frames=self.num_pre_frames,
233
+ num_generate_length=self.num_generate_length,
234
+ audio_feat_win_size=aud_feat_win_size,
235
+ context_info=context_info,
236
+ convert_to_6d=convert_to_6d,
237
+ expression=expression,
238
+ config=self.config,
239
+ am=am,
240
+ am_sr=am_sr,
241
+ whole_video=config.Data.whole_video
242
+ )
243
+ self.complete_data.append(self.dataset[key].complete_data)
244
+ haode = haode + 1
245
+ print("huaide:{}, haode:{}".format(huaide, haode))
246
+ import pickle
247
+
248
+ f = open(self.split+config.Data.pklname, 'wb')
249
+ pickle.dump(self.dataset, f)
250
+ f.close()
251
+ ######################origin load method
252
+
253
+ self.complete_data=np.concatenate(self.complete_data, axis=0)
254
+
255
+ # assert self.complete_data.shape[-1] == (12+21+21)*2
256
+ self.normalize_stats = {}
257
+
258
+ self.data_mean = None
259
+ self.data_std = None
260
+
261
+ def get_dataset(self):
262
+ self.normalize_stats['mean'] = self.data_mean
263
+ self.normalize_stats['std'] = self.data_std
264
+
265
+ for key in list(self.dataset.keys()):
266
+ if self.dataset[key].complete_data.shape[0] < self.num_generate_length:
267
+ continue
268
+ self.dataset[key].num_generate_length = self.num_generate_length
269
+ self.dataset[key].get_dataset(self.normalization, self.normalize_stats, self.split)
270
+ self.all_dataset_list.append(self.dataset[key].all_dataset)
271
+
272
+ if self.split_trans_zero:
273
+ self.trans_dataset = data.ConcatDataset(self.trans_dataset_list)
274
+ self.zero_dataset = data.ConcatDataset(self.zero_dataset_list)
275
+ else:
276
+ self.all_dataset = data.ConcatDataset(self.all_dataset_list)
277
+
278
+
279
+
data_utils/dataset_preprocess.py ADDED
@@ -0,0 +1,170 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import pickle
3
+ from tqdm import tqdm
4
+ import shutil
5
+ import torch
6
+ import numpy as np
7
+ import librosa
8
+ import random
9
+
10
+ speakers = ['seth', 'conan', 'oliver', 'chemistry']
11
+ data_root = "../ExpressiveWholeBodyDatasetv1.0/"
12
+ split = 'train'
13
+
14
+
15
+
16
+ def split_list(full_list,shuffle=False,ratio=0.2):
17
+ n_total = len(full_list)
18
+ offset_0 = int(n_total * ratio)
19
+ offset_1 = int(n_total * ratio * 2)
20
+ if n_total==0 or offset_1<1:
21
+ return [],full_list
22
+ if shuffle:
23
+ random.shuffle(full_list)
24
+ sublist_0 = full_list[:offset_0]
25
+ sublist_1 = full_list[offset_0:offset_1]
26
+ sublist_2 = full_list[offset_1:]
27
+ return sublist_0, sublist_1, sublist_2
28
+
29
+
30
+ def moveto(list, file):
31
+ for f in list:
32
+ before, after = '/'.join(f.split('/')[:-1]), f.split('/')[-1]
33
+ new_path = os.path.join(before, file)
34
+ new_path = os.path.join(new_path, after)
35
+ # os.makedirs(new_path)
36
+ # os.path.isdir(new_path)
37
+ # shutil.move(f, new_path)
38
+
39
+ #转移到新目录
40
+ shutil.copytree(f, new_path)
41
+ #删除原train里的文件
42
+ shutil.rmtree(f)
43
+ return None
44
+
45
+
46
+ def read_pkl(data):
47
+ betas = np.array(data['betas'])
48
+
49
+ jaw_pose = np.array(data['jaw_pose'])
50
+ leye_pose = np.array(data['leye_pose'])
51
+ reye_pose = np.array(data['reye_pose'])
52
+ global_orient = np.array(data['global_orient']).squeeze()
53
+ body_pose = np.array(data['body_pose_axis'])
54
+ left_hand_pose = np.array(data['left_hand_pose'])
55
+ right_hand_pose = np.array(data['right_hand_pose'])
56
+
57
+ full_body = np.concatenate(
58
+ (jaw_pose, leye_pose, reye_pose, global_orient, body_pose, left_hand_pose, right_hand_pose), axis=1)
59
+
60
+ expression = np.array(data['expression'])
61
+ full_body = np.concatenate((full_body, expression), axis=1)
62
+
63
+ if (full_body.shape[0] < 90) or (torch.isnan(torch.from_numpy(full_body)).sum() > 0):
64
+ return 1
65
+ else:
66
+ return 0
67
+
68
+
69
+ for speaker_name in speakers:
70
+ speaker_root = os.path.join(data_root, speaker_name)
71
+
72
+ videos = [v for v in os.listdir(speaker_root)]
73
+ print(videos)
74
+
75
+ haode = huaide = 0
76
+ total_seqs = []
77
+
78
+ for vid in tqdm(videos, desc="Processing training data of {}......".format(speaker_name)):
79
+ # for vid in videos:
80
+ source_vid = vid
81
+ vid_pth = os.path.join(speaker_root, source_vid)
82
+ # vid_pth = os.path.join(speaker_root, source_vid, 'images/half', split)
83
+ t = os.path.join(speaker_root, source_vid, 'test')
84
+ v = os.path.join(speaker_root, source_vid, 'val')
85
+
86
+ # if os.path.exists(t):
87
+ # shutil.rmtree(t)
88
+ # if os.path.exists(v):
89
+ # shutil.rmtree(v)
90
+ try:
91
+ seqs = [s for s in os.listdir(vid_pth)]
92
+ except:
93
+ continue
94
+ # if len(seqs) == 0:
95
+ # shutil.rmtree(os.path.join(speaker_root, source_vid))
96
+ # None
97
+ for s in seqs:
98
+ quality = 0
99
+ total_seqs.append(os.path.join(vid_pth,s))
100
+ seq_root = os.path.join(vid_pth, s)
101
+ key = seq_root # correspond to clip******
102
+ audio_fname = os.path.join(speaker_root, source_vid, s, '%s.wav' % (s))
103
+
104
+ # delete the data without audio or the audio file could not be read
105
+ if os.path.isfile(audio_fname):
106
+ try:
107
+ audio = librosa.load(audio_fname)
108
+ except:
109
+ # print(key)
110
+ shutil.rmtree(key)
111
+ huaide = huaide + 1
112
+ continue
113
+ else:
114
+ huaide = huaide + 1
115
+ # print(key)
116
+ shutil.rmtree(key)
117
+ continue
118
+
119
+ # check motion file
120
+ motion_fname = os.path.join(speaker_root, source_vid, s, '%s.pkl' % (s))
121
+ try:
122
+ f = open(motion_fname, 'rb+')
123
+ except:
124
+ shutil.rmtree(key)
125
+ huaide = huaide + 1
126
+ continue
127
+
128
+ data = pickle.load(f)
129
+ w = read_pkl(data)
130
+ f.close()
131
+ quality = quality + w
132
+
133
+ if w == 1:
134
+ shutil.rmtree(key)
135
+ # print(key)
136
+ huaide = huaide + 1
137
+ continue
138
+
139
+ haode = haode + 1
140
+
141
+ print("huaide:{}, haode:{}, total_seqs:{}".format(huaide, haode, total_seqs.__len__()))
142
+
143
+ for speaker_name in speakers:
144
+ speaker_root = os.path.join(data_root, speaker_name)
145
+
146
+ videos = [v for v in os.listdir(speaker_root)]
147
+ print(videos)
148
+
149
+ haode = huaide = 0
150
+ total_seqs = []
151
+
152
+ for vid in tqdm(videos, desc="Processing training data of {}......".format(speaker_name)):
153
+ # for vid in videos:
154
+ source_vid = vid
155
+ vid_pth = os.path.join(speaker_root, source_vid)
156
+ try:
157
+ seqs = [s for s in os.listdir(vid_pth)]
158
+ except:
159
+ continue
160
+ for s in seqs:
161
+ quality = 0
162
+ total_seqs.append(os.path.join(vid_pth, s))
163
+ print("total_seqs:{}".format(total_seqs.__len__()))
164
+ # split the dataset
165
+ test_list, val_list, train_list = split_list(total_seqs, True, 0.1)
166
+ print(len(test_list), len(val_list), len(train_list))
167
+ moveto(train_list, 'train')
168
+ moveto(test_list, 'test')
169
+ moveto(val_list, 'val')
170
+
data_utils/get_j.py ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import torch
2
+
3
+
4
+ def to3d(poses, config):
5
+ if config.Data.pose.convert_to_6d:
6
+ if config.Data.pose.expression:
7
+ poses_exp = poses[:, -100:]
8
+ poses = poses[:, :-100]
9
+
10
+ poses = poses.reshape(poses.shape[0], -1, 5)
11
+ sin, cos = poses[:, :, 3], poses[:, :, 4]
12
+ pose_angle = torch.atan2(sin, cos)
13
+ poses = (poses[:, :, :3] * pose_angle.unsqueeze(dim=-1)).reshape(poses.shape[0], -1)
14
+
15
+ if config.Data.pose.expression:
16
+ poses = torch.cat([poses, poses_exp], dim=-1)
17
+ return poses
18
+
19
+
20
+ def get_joint(smplx_model, betas, pred):
21
+ joint = smplx_model(betas=betas.repeat(pred.shape[0], 1),
22
+ expression=pred[:, 165:265],
23
+ jaw_pose=pred[:, 0:3],
24
+ leye_pose=pred[:, 3:6],
25
+ reye_pose=pred[:, 6:9],
26
+ global_orient=pred[:, 9:12],
27
+ body_pose=pred[:, 12:75],
28
+ left_hand_pose=pred[:, 75:120],
29
+ right_hand_pose=pred[:, 120:165],
30
+ return_verts=True)['joints']
31
+ return joint
32
+
33
+
34
+ def get_joints(smplx_model, betas, pred):
35
+ if len(pred.shape) == 3:
36
+ B = pred.shape[0]
37
+ x = 4 if B>= 4 else B
38
+ T = pred.shape[1]
39
+ pred = pred.reshape(-1, 265)
40
+ smplx_model.batch_size = L = T * x
41
+
42
+ times = pred.shape[0] // smplx_model.batch_size
43
+ joints = []
44
+ for i in range(times):
45
+ joints.append(get_joint(smplx_model, betas, pred[i*L:(i+1)*L]))
46
+ joints = torch.cat(joints, dim=0)
47
+ joints = joints.reshape(B, T, -1, 3)
48
+ else:
49
+ smplx_model.batch_size = pred.shape[0]
50
+ joints = get_joint(smplx_model, betas, pred)
51
+ return joints
data_utils/hand_component.json ADDED
The diff for this file is too large to render. See raw diff
 
data_utils/lower_body.py ADDED
@@ -0,0 +1,143 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import numpy as np
2
+ import torch
3
+
4
+ lower_pose = torch.tensor(
5
+ [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 3.0747, -0.0158, -0.0152, -1.1826512813568115, 0.23866955935955048,
6
+ 0.15146760642528534, -1.2604516744613647, -0.3160211145877838,
7
+ -0.1603458970785141, 1.1654603481292725, 0.0, 0.0, 1.2521806955337524, 0.041598282754421234, -0.06312154978513718,
8
+ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0])
9
+ lower_pose_stand = torch.tensor([
10
+ 8.9759e-04, 7.1074e-04, -5.9163e-06, 8.9759e-04, 7.1074e-04, -5.9163e-06,
11
+ 3.0747, -0.0158, -0.0152,
12
+ -3.6665e-01, -8.8455e-03, 1.6113e-01, -3.6665e-01, -8.8455e-03, 1.6113e-01,
13
+ -3.9716e-01, -4.0229e-02, -1.2637e-01,
14
+ 7.9163e-01, 6.8519e-02, -1.5091e-01, 7.9163e-01, 6.8519e-02, -1.5091e-01,
15
+ 7.8632e-01, -4.3810e-02, 1.4375e-02,
16
+ -1.0675e-01, 1.2635e-01, 1.6711e-02, -1.0675e-01, 1.2635e-01, 1.6711e-02, ])
17
+ # lower_pose_stand = torch.tensor(
18
+ # [6.4919e-02, 3.3018e-02, 1.7485e-02, 8.9759e-04, 7.1074e-04, -5.9163e-06,
19
+ # 3.0747, -0.0158, -0.0152,
20
+ # -3.3633e+00, -9.3915e-02, 3.0996e-01, -3.6665e-01, -8.8455e-03, 1.6113e-01,
21
+ # 1.1654603481292725, 0.0, 0.0,
22
+ # 4.4167e-01, 6.7183e-03, -3.6379e-03, 7.9163e-01, 6.8519e-02, -1.5091e-01,
23
+ # 0.0, 0.0, 0.0,
24
+ # 2.2910e-02, -2.4797e-02, -5.5657e-03, -1.0675e-01, 1.2635e-01, 1.6711e-02,])
25
+ lower_body = [0, 1, 3, 4, 6, 7, 9, 10]
26
+ count_part = [6, 9, 12, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
27
+ 29, 30, 31, 32, 33, 34, 35, 36, 37,
28
+ 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54]
29
+ fix_index = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
30
+ 29,
31
+ 35, 36, 37, 38, 39, 40, 41, 42, 43, 44,
32
+ 50, 51, 52, 53, 54, 55, 56, 57, 58, 59,
33
+ 65, 66, 67, 68, 69, 70, 71, 72, 73, 74]
34
+ all_index = np.ones(275)
35
+ all_index[fix_index] = 0
36
+ c_index = []
37
+ i = 0
38
+ for num in all_index:
39
+ if num == 1:
40
+ c_index.append(i)
41
+ i = i + 1
42
+ c_index = np.asarray(c_index)
43
+
44
+ fix_index_3d = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,
45
+ 21, 22, 23, 24, 25, 26,
46
+ 30, 31, 32, 33, 34, 35,
47
+ 45, 46, 47, 48, 49, 50]
48
+ all_index_3d = np.ones(165)
49
+ all_index_3d[fix_index_3d] = 0
50
+ c_index_3d = []
51
+ i = 0
52
+ for num in all_index_3d:
53
+ if num == 1:
54
+ c_index_3d.append(i)
55
+ i = i + 1
56
+ c_index_3d = np.asarray(c_index_3d)
57
+
58
+ c_index_6d = []
59
+ i = 0
60
+ for num in all_index_3d:
61
+ if num == 1:
62
+ c_index_6d.append(2*i)
63
+ c_index_6d.append(2 * i + 1)
64
+ i = i + 1
65
+ c_index_6d = np.asarray(c_index_6d)
66
+
67
+
68
+ def part2full(input, stand=False):
69
+ if stand:
70
+ # lp = lower_pose_stand.unsqueeze(dim=0).repeat(input.shape[0], 1).to(input.device)
71
+ lp = torch.zeros_like(lower_pose)
72
+ lp[6:9] = torch.tensor([3.0747, -0.0158, -0.0152])
73
+ lp = lp.unsqueeze(dim=0).repeat(input.shape[0], 1).to(input.device)
74
+ else:
75
+ lp = lower_pose.unsqueeze(dim=0).repeat(input.shape[0], 1).to(input.device)
76
+
77
+ input = torch.cat([input[:, :3],
78
+ lp[:, :15],
79
+ input[:, 3:6],
80
+ lp[:, 15:21],
81
+ input[:, 6:9],
82
+ lp[:, 21:27],
83
+ input[:, 9:12],
84
+ lp[:, 27:],
85
+ input[:, 12:]]
86
+ , dim=1)
87
+ return input
88
+
89
+
90
+ def pred2poses(input, gt):
91
+ input = torch.cat([input[:, :3],
92
+ gt[0:1, 3:18].repeat(input.shape[0], 1),
93
+ input[:, 3:6],
94
+ gt[0:1, 21:27].repeat(input.shape[0], 1),
95
+ input[:, 6:9],
96
+ gt[0:1, 30:36].repeat(input.shape[0], 1),
97
+ input[:, 9:12],
98
+ gt[0:1, 39:45].repeat(input.shape[0], 1),
99
+ input[:, 12:]]
100
+ , dim=1)
101
+ return input
102
+
103
+
104
+ def poses2poses(input, gt):
105
+ input = torch.cat([input[:, :3],
106
+ gt[0:1, 3:18].repeat(input.shape[0], 1),
107
+ input[:, 18:21],
108
+ gt[0:1, 21:27].repeat(input.shape[0], 1),
109
+ input[:, 27:30],
110
+ gt[0:1, 30:36].repeat(input.shape[0], 1),
111
+ input[:, 36:39],
112
+ gt[0:1, 39:45].repeat(input.shape[0], 1),
113
+ input[:, 45:]]
114
+ , dim=1)
115
+ return input
116
+
117
+ def poses2pred(input, stand=False):
118
+ if stand:
119
+ lp = lower_pose_stand.unsqueeze(dim=0).repeat(input.shape[0], 1).to(input.device)
120
+ # lp = torch.zeros_like(lower_pose).unsqueeze(dim=0).repeat(input.shape[0], 1).to(input.device)
121
+ else:
122
+ lp = lower_pose.unsqueeze(dim=0).repeat(input.shape[0], 1).to(input.device)
123
+ input = torch.cat([input[:, :3],
124
+ lp[:, :15],
125
+ input[:, 18:21],
126
+ lp[:, 15:21],
127
+ input[:, 27:30],
128
+ lp[:, 21:27],
129
+ input[:, 36:39],
130
+ lp[:, 27:],
131
+ input[:, 45:]]
132
+ , dim=1)
133
+ return input
134
+
135
+
136
+ rearrange = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21]\
137
+ # ,22, 23, 24, 25, 40, 26, 41,
138
+ # 27, 42, 28, 43, 29, 44, 30, 45, 31, 46, 32, 47, 33, 48, 34, 49, 35, 50, 36, 51, 37, 52, 38, 53, 39, 54, 55,
139
+ # 57, 56, 59, 58, 60, 63, 61, 64, 62, 65, 66, 71, 67, 72, 68, 73, 69, 74, 70, 75]
140
+
141
+ symmetry = [0, 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1]#, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
142
+ # 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
143
+ # 1, 1, 1, 1, 1, 1]
data_utils/mesh_dataset.py ADDED
@@ -0,0 +1,348 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import pickle
2
+ import sys
3
+ import os
4
+
5
+ sys.path.append(os.getcwd())
6
+
7
+ import json
8
+ from glob import glob
9
+ from data_utils.utils import *
10
+ import torch.utils.data as data
11
+ from data_utils.consts import speaker_id
12
+ from data_utils.lower_body import count_part
13
+ import random
14
+ from data_utils.rotation_conversion import axis_angle_to_matrix, matrix_to_rotation_6d
15
+
16
+ with open('data_utils/hand_component.json') as file_obj:
17
+ comp = json.load(file_obj)
18
+ left_hand_c = np.asarray(comp['left'])
19
+ right_hand_c = np.asarray(comp['right'])
20
+
21
+
22
+ def to3d(data):
23
+ left_hand_pose = np.einsum('bi,ij->bj', data[:, 75:87], left_hand_c[:12, :])
24
+ right_hand_pose = np.einsum('bi,ij->bj', data[:, 87:99], right_hand_c[:12, :])
25
+ data = np.concatenate((data[:, :75], left_hand_pose, right_hand_pose), axis=-1)
26
+ return data
27
+
28
+
29
+ class SmplxDataset():
30
+ '''
31
+ creat a dataset for every segment and concat.
32
+ '''
33
+
34
+ def __init__(self,
35
+ data_root,
36
+ speaker,
37
+ motion_fn,
38
+ audio_fn,
39
+ audio_sr,
40
+ fps,
41
+ feat_method='mel_spec',
42
+ audio_feat_dim=64,
43
+ audio_feat_win_size=None,
44
+
45
+ train=True,
46
+ load_all=False,
47
+ split_trans_zero=False,
48
+ limbscaling=False,
49
+ num_frames=25,
50
+ num_pre_frames=25,
51
+ num_generate_length=25,
52
+ context_info=False,
53
+ convert_to_6d=False,
54
+ expression=False,
55
+ config=None,
56
+ am=None,
57
+ am_sr=None,
58
+ whole_video=False
59
+ ):
60
+
61
+ self.data_root = data_root
62
+ self.speaker = speaker
63
+
64
+ self.feat_method = feat_method
65
+ self.audio_fn = audio_fn
66
+ self.audio_sr = audio_sr
67
+ self.fps = fps
68
+ self.audio_feat_dim = audio_feat_dim
69
+ self.audio_feat_win_size = audio_feat_win_size
70
+ self.context_info = context_info # for aud feat
71
+ self.convert_to_6d = convert_to_6d
72
+ self.expression = expression
73
+
74
+ self.train = train
75
+ self.load_all = load_all
76
+ self.split_trans_zero = split_trans_zero
77
+ self.limbscaling = limbscaling
78
+ self.num_frames = num_frames
79
+ self.num_pre_frames = num_pre_frames
80
+ self.num_generate_length = num_generate_length
81
+ # print('num_generate_length ', self.num_generate_length)
82
+
83
+ self.config = config
84
+ self.am_sr = am_sr
85
+ self.whole_video = whole_video
86
+ load_mode = self.config.dataset_load_mode
87
+
88
+ if load_mode == 'pickle':
89
+ raise NotImplementedError
90
+
91
+ elif load_mode == 'csv':
92
+ import pickle
93
+ with open(data_root, 'rb') as f:
94
+ u = pickle._Unpickler(f)
95
+ data = u.load()
96
+ self.data = data[0]
97
+ if self.load_all:
98
+ self._load_npz_all()
99
+
100
+ elif load_mode == 'json':
101
+ self.annotations = glob(data_root + '/*pkl')
102
+ if len(self.annotations) == 0:
103
+ raise FileNotFoundError(data_root + ' are empty')
104
+ self.annotations = sorted(self.annotations)
105
+ self.img_name_list = self.annotations
106
+
107
+ if self.load_all:
108
+ self._load_them_all(am, am_sr, motion_fn)
109
+
110
+ def _load_npz_all(self):
111
+ self.loaded_data = {}
112
+ self.complete_data = []
113
+ data = self.data
114
+ shape = data['body_pose_axis'].shape[0]
115
+ self.betas = data['betas']
116
+ self.img_name_list = []
117
+ for index in range(shape):
118
+ img_name = f'{index:6d}'
119
+ self.img_name_list.append(img_name)
120
+
121
+ jaw_pose = data['jaw_pose'][index]
122
+ leye_pose = data['leye_pose'][index]
123
+ reye_pose = data['reye_pose'][index]
124
+ global_orient = data['global_orient'][index]
125
+ body_pose = data['body_pose_axis'][index]
126
+ left_hand_pose = data['left_hand_pose'][index]
127
+ right_hand_pose = data['right_hand_pose'][index]
128
+
129
+ full_body = np.concatenate(
130
+ (jaw_pose, leye_pose, reye_pose, global_orient, body_pose, left_hand_pose, right_hand_pose))
131
+ assert full_body.shape[0] == 99
132
+ if self.convert_to_6d:
133
+ full_body = to3d(full_body)
134
+ full_body = torch.from_numpy(full_body)
135
+ full_body = matrix_to_rotation_6d(axis_angle_to_matrix(full_body))
136
+ full_body = np.asarray(full_body)
137
+ if self.expression:
138
+ expression = data['expression'][index]
139
+ full_body = np.concatenate((full_body, expression))
140
+ # full_body = np.concatenate((full_body, non_zero))
141
+ else:
142
+ full_body = to3d(full_body)
143
+ if self.expression:
144
+ expression = data['expression'][index]
145
+ full_body = np.concatenate((full_body, expression))
146
+
147
+ self.loaded_data[img_name] = full_body.reshape(-1)
148
+ self.complete_data.append(full_body.reshape(-1))
149
+
150
+ self.complete_data = np.array(self.complete_data)
151
+
152
+ if self.audio_feat_win_size is not None:
153
+ self.audio_feat = get_mfcc_old(self.audio_fn).transpose(1, 0)
154
+ # print(self.audio_feat.shape)
155
+ else:
156
+ if self.feat_method == 'mel_spec':
157
+ self.audio_feat = get_melspec(self.audio_fn, fps=self.fps, sr=self.audio_sr, n_mels=self.audio_feat_dim)
158
+ elif self.feat_method == 'mfcc':
159
+ self.audio_feat = get_mfcc(self.audio_fn,
160
+ smlpx=True,
161
+ sr=self.audio_sr,
162
+ n_mfcc=self.audio_feat_dim,
163
+ win_size=self.audio_feat_win_size
164
+ )
165
+
166
+ def _load_them_all(self, am, am_sr, motion_fn):
167
+ self.loaded_data = {}
168
+ self.complete_data = []
169
+ f = open(motion_fn, 'rb+')
170
+ data = pickle.load(f)
171
+
172
+ self.betas = np.array(data['betas'])
173
+
174
+ jaw_pose = np.array(data['jaw_pose'])
175
+ leye_pose = np.array(data['leye_pose'])
176
+ reye_pose = np.array(data['reye_pose'])
177
+ global_orient = np.array(data['global_orient']).squeeze()
178
+ body_pose = np.array(data['body_pose_axis'])
179
+ left_hand_pose = np.array(data['left_hand_pose'])
180
+ right_hand_pose = np.array(data['right_hand_pose'])
181
+
182
+ full_body = np.concatenate(
183
+ (jaw_pose, leye_pose, reye_pose, global_orient, body_pose, left_hand_pose, right_hand_pose), axis=1)
184
+ assert full_body.shape[1] == 99
185
+
186
+
187
+ if self.convert_to_6d:
188
+ full_body = to3d(full_body)
189
+ full_body = torch.from_numpy(full_body)
190
+ full_body = matrix_to_rotation_6d(axis_angle_to_matrix(full_body.reshape(-1, 55, 3))).reshape(-1, 330)
191
+ full_body = np.asarray(full_body)
192
+ if self.expression:
193
+ expression = np.array(data['expression'])
194
+ full_body = np.concatenate((full_body, expression), axis=1)
195
+
196
+ else:
197
+ full_body = to3d(full_body)
198
+ expression = np.array(data['expression'])
199
+ full_body = np.concatenate((full_body, expression), axis=1)
200
+
201
+ self.complete_data = full_body
202
+ self.complete_data = np.array(self.complete_data)
203
+
204
+ if self.audio_feat_win_size is not None:
205
+ self.audio_feat = get_mfcc_old(self.audio_fn).transpose(1, 0)
206
+ else:
207
+ # if self.feat_method == 'mel_spec':
208
+ # self.audio_feat = get_melspec(self.audio_fn, fps=self.fps, sr=self.audio_sr, n_mels=self.audio_feat_dim)
209
+ # elif self.feat_method == 'mfcc':
210
+ self.audio_feat = get_mfcc_ta(self.audio_fn,
211
+ smlpx=True,
212
+ fps=30,
213
+ sr=self.audio_sr,
214
+ n_mfcc=self.audio_feat_dim,
215
+ win_size=self.audio_feat_win_size,
216
+ type=self.feat_method,
217
+ am=am,
218
+ am_sr=am_sr,
219
+ encoder_choice=self.config.Model.encoder_choice,
220
+ )
221
+ # with open(audio_file, 'w', encoding='utf-8') as file:
222
+ # file.write(json.dumps(self.audio_feat.__array__().tolist(), indent=0, ensure_ascii=False))
223
+
224
+ def get_dataset(self, normalization=False, normalize_stats=None, split='train'):
225
+
226
+ class __Worker__(data.Dataset):
227
+ def __init__(child, index_list, normalization, normalize_stats, split='train') -> None:
228
+ super().__init__()
229
+ child.index_list = index_list
230
+ child.normalization = normalization
231
+ child.normalize_stats = normalize_stats
232
+ child.split = split
233
+
234
+ def __getitem__(child, index):
235
+ num_generate_length = self.num_generate_length
236
+ num_pre_frames = self.num_pre_frames
237
+ seq_len = num_generate_length + num_pre_frames
238
+ # print(num_generate_length)
239
+
240
+ index = child.index_list[index]
241
+ index_new = index + random.randrange(0, 5, 3)
242
+ if index_new + seq_len > self.complete_data.shape[0]:
243
+ index_new = index
244
+ index = index_new
245
+
246
+ if child.split in ['val', 'pre', 'test'] or self.whole_video:
247
+ index = 0
248
+ seq_len = self.complete_data.shape[0]
249
+ seq_data = []
250
+ assert index + seq_len <= self.complete_data.shape[0]
251
+ # print(seq_len)
252
+ seq_data = self.complete_data[index:(index + seq_len), :]
253
+ seq_data = np.array(seq_data)
254
+
255
+ '''
256
+ audio feature,
257
+ '''
258
+ if not self.context_info:
259
+ if not self.whole_video:
260
+ audio_feat = self.audio_feat[index:index + seq_len, ...]
261
+ if audio_feat.shape[0] < seq_len:
262
+ audio_feat = np.pad(audio_feat, [[0, seq_len - audio_feat.shape[0]], [0, 0]],
263
+ mode='reflect')
264
+
265
+ assert audio_feat.shape[0] == seq_len and audio_feat.shape[1] == self.audio_feat_dim
266
+ else:
267
+ audio_feat = self.audio_feat
268
+
269
+ else: # including feature and history
270
+ if self.audio_feat_win_size is None:
271
+ audio_feat = self.audio_feat[index:index + seq_len + num_pre_frames, ...]
272
+ if audio_feat.shape[0] < seq_len + num_pre_frames:
273
+ audio_feat = np.pad(audio_feat,
274
+ [[0, seq_len + self.num_frames - audio_feat.shape[0]], [0, 0]],
275
+ mode='constant')
276
+
277
+ assert audio_feat.shape[0] == self.num_frames + seq_len and audio_feat.shape[
278
+ 1] == self.audio_feat_dim
279
+
280
+ if child.normalization:
281
+ data_mean = child.normalize_stats['mean'].reshape(1, -1)
282
+ data_std = child.normalize_stats['std'].reshape(1, -1)
283
+ seq_data[:, :330] = (seq_data[:, :330] - data_mean) / data_std
284
+ if child.split in['train', 'test']:
285
+ if self.convert_to_6d:
286
+ if self.expression:
287
+ data_sample = {
288
+ 'poses': seq_data[:, :330].astype(np.float).transpose(1, 0),
289
+ 'expression': seq_data[:, 330:].astype(np.float).transpose(1, 0),
290
+ # 'nzero': seq_data[:, 375:].astype(np.float).transpose(1, 0),
291
+ 'aud_feat': audio_feat.astype(np.float).transpose(1, 0),
292
+ 'speaker': speaker_id[self.speaker],
293
+ 'betas': self.betas,
294
+ 'aud_file': self.audio_fn,
295
+ }
296
+ else:
297
+ data_sample = {
298
+ 'poses': seq_data[:, :330].astype(np.float).transpose(1, 0),
299
+ 'nzero': seq_data[:, 330:].astype(np.float).transpose(1, 0),
300
+ 'aud_feat': audio_feat.astype(np.float).transpose(1, 0),
301
+ 'speaker': speaker_id[self.speaker],
302
+ 'betas': self.betas
303
+ }
304
+ else:
305
+ if self.expression:
306
+ data_sample = {
307
+ 'poses': seq_data[:, :165].astype(np.float).transpose(1, 0),
308
+ 'expression': seq_data[:, 165:].astype(np.float).transpose(1, 0),
309
+ 'aud_feat': audio_feat.astype(np.float).transpose(1, 0),
310
+ # 'wv2_feat': wv2_feat.astype(np.float).transpose(1, 0),
311
+ 'speaker': speaker_id[self.speaker],
312
+ 'aud_file': self.audio_fn,
313
+ 'betas': self.betas
314
+ }
315
+ else:
316
+ data_sample = {
317
+ 'poses': seq_data.astype(np.float).transpose(1, 0),
318
+ 'aud_feat': audio_feat.astype(np.float).transpose(1, 0),
319
+ 'speaker': speaker_id[self.speaker],
320
+ 'betas': self.betas
321
+ }
322
+ return data_sample
323
+ else:
324
+ data_sample = {
325
+ 'poses': seq_data[:, :330].astype(np.float).transpose(1, 0),
326
+ 'expression': seq_data[:, 330:].astype(np.float).transpose(1, 0),
327
+ # 'nzero': seq_data[:, 325:].astype(np.float).transpose(1, 0),
328
+ 'aud_feat': audio_feat.astype(np.float).transpose(1, 0),
329
+ 'aud_file': self.audio_fn,
330
+ 'speaker': speaker_id[self.speaker],
331
+ 'betas': self.betas
332
+ }
333
+ return data_sample
334
+ def __len__(child):
335
+ return len(child.index_list)
336
+
337
+ if split == 'train':
338
+ index_list = list(
339
+ range(0, min(self.complete_data.shape[0], self.audio_feat.shape[0]) - self.num_generate_length - self.num_pre_frames,
340
+ 6))
341
+ elif split in ['val', 'test']:
342
+ index_list = list([0])
343
+ if self.whole_video:
344
+ index_list = list([0])
345
+ self.all_dataset = __Worker__(index_list, normalization, normalize_stats, split)
346
+
347
+ def __len__(self):
348
+ return len(self.img_name_list)
data_utils/rotation_conversion.py ADDED
@@ -0,0 +1,551 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) Facebook, Inc. and its affiliates. All rights reserved.
2
+ # Check PYTORCH3D_LICENCE before use
3
+
4
+ import functools
5
+ from typing import Optional
6
+
7
+ import torch
8
+ import torch.nn.functional as F
9
+
10
+
11
+ """
12
+ The transformation matrices returned from the functions in this file assume
13
+ the points on which the transformation will be applied are column vectors.
14
+ i.e. the R matrix is structured as
15
+
16
+ R = [
17
+ [Rxx, Rxy, Rxz],
18
+ [Ryx, Ryy, Ryz],
19
+ [Rzx, Rzy, Rzz],
20
+ ] # (3, 3)
21
+
22
+ This matrix can be applied to column vectors by post multiplication
23
+ by the points e.g.
24
+
25
+ points = [[0], [1], [2]] # (3 x 1) xyz coordinates of a point
26
+ transformed_points = R * points
27
+
28
+ To apply the same matrix to points which are row vectors, the R matrix
29
+ can be transposed and pre multiplied by the points:
30
+
31
+ e.g.
32
+ points = [[0, 1, 2]] # (1 x 3) xyz coordinates of a point
33
+ transformed_points = points * R.transpose(1, 0)
34
+ """
35
+
36
+
37
+ def quaternion_to_matrix(quaternions):
38
+ """
39
+ Convert rotations given as quaternions to rotation matrices.
40
+
41
+ Args:
42
+ quaternions: quaternions with real part first,
43
+ as tensor of shape (..., 4).
44
+
45
+ Returns:
46
+ Rotation matrices as tensor of shape (..., 3, 3).
47
+ """
48
+ r, i, j, k = torch.unbind(quaternions, -1)
49
+ two_s = 2.0 / (quaternions * quaternions).sum(-1)
50
+
51
+ o = torch.stack(
52
+ (
53
+ 1 - two_s * (j * j + k * k),
54
+ two_s * (i * j - k * r),
55
+ two_s * (i * k + j * r),
56
+ two_s * (i * j + k * r),
57
+ 1 - two_s * (i * i + k * k),
58
+ two_s * (j * k - i * r),
59
+ two_s * (i * k - j * r),
60
+ two_s * (j * k + i * r),
61
+ 1 - two_s * (i * i + j * j),
62
+ ),
63
+ -1,
64
+ )
65
+ return o.reshape(quaternions.shape[:-1] + (3, 3))
66
+
67
+
68
+ def _copysign(a, b):
69
+ """
70
+ Return a tensor where each element has the absolute value taken from the,
71
+ corresponding element of a, with sign taken from the corresponding
72
+ element of b. This is like the standard copysign floating-point operation,
73
+ but is not careful about negative 0 and NaN.
74
+
75
+ Args:
76
+ a: source tensor.
77
+ b: tensor whose signs will be used, of the same shape as a.
78
+
79
+ Returns:
80
+ Tensor of the same shape as a with the signs of b.
81
+ """
82
+ signs_differ = (a < 0) != (b < 0)
83
+ return torch.where(signs_differ, -a, a)
84
+
85
+
86
+ def _sqrt_positive_part(x):
87
+ """
88
+ Returns torch.sqrt(torch.max(0, x))
89
+ but with a zero subgradient where x is 0.
90
+ """
91
+ ret = torch.zeros_like(x)
92
+ positive_mask = x > 0
93
+ ret[positive_mask] = torch.sqrt(x[positive_mask])
94
+ return ret
95
+
96
+
97
+ def matrix_to_quaternion(matrix):
98
+ """
99
+ Convert rotations given as rotation matrices to quaternions.
100
+
101
+ Args:
102
+ matrix: Rotation matrices as tensor of shape (..., 3, 3).
103
+
104
+ Returns:
105
+ quaternions with real part first, as tensor of shape (..., 4).
106
+ """
107
+ if matrix.size(-1) != 3 or matrix.size(-2) != 3:
108
+ raise ValueError(f"Invalid rotation matrix shape f{matrix.shape}.")
109
+ m00 = matrix[..., 0, 0]
110
+ m11 = matrix[..., 1, 1]
111
+ m22 = matrix[..., 2, 2]
112
+ o0 = 0.5 * _sqrt_positive_part(1 + m00 + m11 + m22)
113
+ x = 0.5 * _sqrt_positive_part(1 + m00 - m11 - m22)
114
+ y = 0.5 * _sqrt_positive_part(1 - m00 + m11 - m22)
115
+ z = 0.5 * _sqrt_positive_part(1 - m00 - m11 + m22)
116
+ o1 = _copysign(x, matrix[..., 2, 1] - matrix[..., 1, 2])
117
+ o2 = _copysign(y, matrix[..., 0, 2] - matrix[..., 2, 0])
118
+ o3 = _copysign(z, matrix[..., 1, 0] - matrix[..., 0, 1])
119
+ return torch.stack((o0, o1, o2, o3), -1)
120
+
121
+
122
+ def _axis_angle_rotation(axis: str, angle):
123
+ """
124
+ Return the rotation matrices for one of the rotations about an axis
125
+ of which Euler angles describe, for each value of the angle given.
126
+
127
+ Args:
128
+ axis: Axis label "X" or "Y or "Z".
129
+ angle: any shape tensor of Euler angles in radians
130
+
131
+ Returns:
132
+ Rotation matrices as tensor of shape (..., 3, 3).
133
+ """
134
+
135
+ cos = torch.cos(angle)
136
+ sin = torch.sin(angle)
137
+ one = torch.ones_like(angle)
138
+ zero = torch.zeros_like(angle)
139
+
140
+ if axis == "X":
141
+ R_flat = (one, zero, zero, zero, cos, -sin, zero, sin, cos)
142
+ if axis == "Y":
143
+ R_flat = (cos, zero, sin, zero, one, zero, -sin, zero, cos)
144
+ if axis == "Z":
145
+ R_flat = (cos, -sin, zero, sin, cos, zero, zero, zero, one)
146
+
147
+ return torch.stack(R_flat, -1).reshape(angle.shape + (3, 3))
148
+
149
+
150
+ def euler_angles_to_matrix(euler_angles, convention: str):
151
+ """
152
+ Convert rotations given as Euler angles in radians to rotation matrices.
153
+
154
+ Args:
155
+ euler_angles: Euler angles in radians as tensor of shape (..., 3).
156
+ convention: Convention string of three uppercase letters from
157
+ {"X", "Y", and "Z"}.
158
+
159
+ Returns:
160
+ Rotation matrices as tensor of shape (..., 3, 3).
161
+ """
162
+ if euler_angles.dim() == 0 or euler_angles.shape[-1] != 3:
163
+ raise ValueError("Invalid input euler angles.")
164
+ if len(convention) != 3:
165
+ raise ValueError("Convention must have 3 letters.")
166
+ if convention[1] in (convention[0], convention[2]):
167
+ raise ValueError(f"Invalid convention {convention}.")
168
+ for letter in convention:
169
+ if letter not in ("X", "Y", "Z"):
170
+ raise ValueError(f"Invalid letter {letter} in convention string.")
171
+ matrices = map(_axis_angle_rotation, convention, torch.unbind(euler_angles, -1))
172
+ return functools.reduce(torch.matmul, matrices)
173
+
174
+
175
+ def _angle_from_tan(
176
+ axis: str, other_axis: str, data, horizontal: bool, tait_bryan: bool
177
+ ):
178
+ """
179
+ Extract the first or third Euler angle from the two members of
180
+ the matrix which are positive constant times its sine and cosine.
181
+
182
+ Args:
183
+ axis: Axis label "X" or "Y or "Z" for the angle we are finding.
184
+ other_axis: Axis label "X" or "Y or "Z" for the middle axis in the
185
+ convention.
186
+ data: Rotation matrices as tensor of shape (..., 3, 3).
187
+ horizontal: Whether we are looking for the angle for the third axis,
188
+ which means the relevant entries are in the same row of the
189
+ rotation matrix. If not, they are in the same column.
190
+ tait_bryan: Whether the first and third axes in the convention differ.
191
+
192
+ Returns:
193
+ Euler Angles in radians for each matrix in data as a tensor
194
+ of shape (...).
195
+ """
196
+
197
+ i1, i2 = {"X": (2, 1), "Y": (0, 2), "Z": (1, 0)}[axis]
198
+ if horizontal:
199
+ i2, i1 = i1, i2
200
+ even = (axis + other_axis) in ["XY", "YZ", "ZX"]
201
+ if horizontal == even:
202
+ return torch.atan2(data[..., i1], data[..., i2])
203
+ if tait_bryan:
204
+ return torch.atan2(-data[..., i2], data[..., i1])
205
+ return torch.atan2(data[..., i2], -data[..., i1])
206
+
207
+
208
+ def _index_from_letter(letter: str):
209
+ if letter == "X":
210
+ return 0
211
+ if letter == "Y":
212
+ return 1
213
+ if letter == "Z":
214
+ return 2
215
+
216
+
217
+ def matrix_to_euler_angles(matrix, convention: str):
218
+ """
219
+ Convert rotations given as rotation matrices to Euler angles in radians.
220
+
221
+ Args:
222
+ matrix: Rotation matrices as tensor of shape (..., 3, 3).
223
+ convention: Convention string of three uppercase letters.
224
+
225
+ Returns:
226
+ Euler angles in radians as tensor of shape (..., 3).
227
+ """
228
+ if len(convention) != 3:
229
+ raise ValueError("Convention must have 3 letters.")
230
+ if convention[1] in (convention[0], convention[2]):
231
+ raise ValueError(f"Invalid convention {convention}.")
232
+ for letter in convention:
233
+ if letter not in ("X", "Y", "Z"):
234
+ raise ValueError(f"Invalid letter {letter} in convention string.")
235
+ if matrix.size(-1) != 3 or matrix.size(-2) != 3:
236
+ raise ValueError(f"Invalid rotation matrix shape f{matrix.shape}.")
237
+ i0 = _index_from_letter(convention[0])
238
+ i2 = _index_from_letter(convention[2])
239
+ tait_bryan = i0 != i2
240
+ if tait_bryan:
241
+ central_angle = torch.asin(
242
+ matrix[..., i0, i2] * (-1.0 if i0 - i2 in [-1, 2] else 1.0)
243
+ )
244
+ else:
245
+ central_angle = torch.acos(matrix[..., i0, i0])
246
+
247
+ o = (
248
+ _angle_from_tan(
249
+ convention[0], convention[1], matrix[..., i2], False, tait_bryan
250
+ ),
251
+ central_angle,
252
+ _angle_from_tan(
253
+ convention[2], convention[1], matrix[..., i0, :], True, tait_bryan
254
+ ),
255
+ )
256
+ return torch.stack(o, -1)
257
+
258
+
259
+ def random_quaternions(
260
+ n: int, dtype: Optional[torch.dtype] = None, device=None, requires_grad=False
261
+ ):
262
+ """
263
+ Generate random quaternions representing rotations,
264
+ i.e. versors with nonnegative real part.
265
+
266
+ Args:
267
+ n: Number of quaternions in a batch to return.
268
+ dtype: Type to return.
269
+ device: Desired device of returned tensor. Default:
270
+ uses the current device for the default tensor type.
271
+ requires_grad: Whether the resulting tensor should have the gradient
272
+ flag set.
273
+
274
+ Returns:
275
+ Quaternions as tensor of shape (N, 4).
276
+ """
277
+ o = torch.randn((n, 4), dtype=dtype, device=device, requires_grad=requires_grad)
278
+ s = (o * o).sum(1)
279
+ o = o / _copysign(torch.sqrt(s), o[:, 0])[:, None]
280
+ return o
281
+
282
+
283
+ def random_rotations(
284
+ n: int, dtype: Optional[torch.dtype] = None, device=None, requires_grad=False
285
+ ):
286
+ """
287
+ Generate random rotations as 3x3 rotation matrices.
288
+
289
+ Args:
290
+ n: Number of rotation matrices in a batch to return.
291
+ dtype: Type to return.
292
+ device: Device of returned tensor. Default: if None,
293
+ uses the current device for the default tensor type.
294
+ requires_grad: Whether the resulting tensor should have the gradient
295
+ flag set.
296
+
297
+ Returns:
298
+ Rotation matrices as tensor of shape (n, 3, 3).
299
+ """
300
+ quaternions = random_quaternions(
301
+ n, dtype=dtype, device=device, requires_grad=requires_grad
302
+ )
303
+ return quaternion_to_matrix(quaternions)
304
+
305
+
306
+ def random_rotation(
307
+ dtype: Optional[torch.dtype] = None, device=None, requires_grad=False
308
+ ):
309
+ """
310
+ Generate a single random 3x3 rotation matrix.
311
+
312
+ Args:
313
+ dtype: Type to return
314
+ device: Device of returned tensor. Default: if None,
315
+ uses the current device for the default tensor type
316
+ requires_grad: Whether the resulting tensor should have the gradient
317
+ flag set
318
+
319
+ Returns:
320
+ Rotation matrix as tensor of shape (3, 3).
321
+ """
322
+ return random_rotations(1, dtype, device, requires_grad)[0]
323
+
324
+
325
+ def standardize_quaternion(quaternions):
326
+ """
327
+ Convert a unit quaternion to a standard form: one in which the real
328
+ part is non negative.
329
+
330
+ Args:
331
+ quaternions: Quaternions with real part first,
332
+ as tensor of shape (..., 4).
333
+
334
+ Returns:
335
+ Standardized quaternions as tensor of shape (..., 4).
336
+ """
337
+ return torch.where(quaternions[..., 0:1] < 0, -quaternions, quaternions)
338
+
339
+
340
+ def quaternion_raw_multiply(a, b):
341
+ """
342
+ Multiply two quaternions.
343
+ Usual torch rules for broadcasting apply.
344
+
345
+ Args:
346
+ a: Quaternions as tensor of shape (..., 4), real part first.
347
+ b: Quaternions as tensor of shape (..., 4), real part first.
348
+
349
+ Returns:
350
+ The product of a and b, a tensor of quaternions shape (..., 4).
351
+ """
352
+ aw, ax, ay, az = torch.unbind(a, -1)
353
+ bw, bx, by, bz = torch.unbind(b, -1)
354
+ ow = aw * bw - ax * bx - ay * by - az * bz
355
+ ox = aw * bx + ax * bw + ay * bz - az * by
356
+ oy = aw * by - ax * bz + ay * bw + az * bx
357
+ oz = aw * bz + ax * by - ay * bx + az * bw
358
+ return torch.stack((ow, ox, oy, oz), -1)
359
+
360
+
361
+ def quaternion_multiply(a, b):
362
+ """
363
+ Multiply two quaternions representing rotations, returning the quaternion
364
+ representing their composition, i.e. the versor with nonnegative real part.
365
+ Usual torch rules for broadcasting apply.
366
+
367
+ Args:
368
+ a: Quaternions as tensor of shape (..., 4), real part first.
369
+ b: Quaternions as tensor of shape (..., 4), real part first.
370
+
371
+ Returns:
372
+ The product of a and b, a tensor of quaternions of shape (..., 4).
373
+ """
374
+ ab = quaternion_raw_multiply(a, b)
375
+ return standardize_quaternion(ab)
376
+
377
+
378
+ def quaternion_invert(quaternion):
379
+ """
380
+ Given a quaternion representing rotation, get the quaternion representing
381
+ its inverse.
382
+
383
+ Args:
384
+ quaternion: Quaternions as tensor of shape (..., 4), with real part
385
+ first, which must be versors (unit quaternions).
386
+
387
+ Returns:
388
+ The inverse, a tensor of quaternions of shape (..., 4).
389
+ """
390
+
391
+ return quaternion * quaternion.new_tensor([1, -1, -1, -1])
392
+
393
+
394
+ def quaternion_apply(quaternion, point):
395
+ """
396
+ Apply the rotation given by a quaternion to a 3D point.
397
+ Usual torch rules for broadcasting apply.
398
+
399
+ Args:
400
+ quaternion: Tensor of quaternions, real part first, of shape (..., 4).
401
+ point: Tensor of 3D points of shape (..., 3).
402
+
403
+ Returns:
404
+ Tensor of rotated points of shape (..., 3).
405
+ """
406
+ if point.size(-1) != 3:
407
+ raise ValueError(f"Points are not in 3D, f{point.shape}.")
408
+ real_parts = point.new_zeros(point.shape[:-1] + (1,))
409
+ point_as_quaternion = torch.cat((real_parts, point), -1)
410
+ out = quaternion_raw_multiply(
411
+ quaternion_raw_multiply(quaternion, point_as_quaternion),
412
+ quaternion_invert(quaternion),
413
+ )
414
+ return out[..., 1:]
415
+
416
+
417
+ def axis_angle_to_matrix(axis_angle):
418
+ """
419
+ Convert rotations given as axis/angle to rotation matrices.
420
+
421
+ Args:
422
+ axis_angle: Rotations given as a vector in axis angle form,
423
+ as a tensor of shape (..., 3), where the magnitude is
424
+ the angle turned anticlockwise in radians around the
425
+ vector's direction.
426
+
427
+ Returns:
428
+ Rotation matrices as tensor of shape (..., 3, 3).
429
+ """
430
+ return quaternion_to_matrix(axis_angle_to_quaternion(axis_angle))
431
+
432
+
433
+ def matrix_to_axis_angle(matrix):
434
+ """
435
+ Convert rotations given as rotation matrices to axis/angle.
436
+
437
+ Args:
438
+ matrix: Rotation matrices as tensor of shape (..., 3, 3).
439
+
440
+ Returns:
441
+ Rotations given as a vector in axis angle form, as a tensor
442
+ of shape (..., 3), where the magnitude is the angle
443
+ turned anticlockwise in radians around the vector's
444
+ direction.
445
+ """
446
+ return quaternion_to_axis_angle(matrix_to_quaternion(matrix))
447
+
448
+
449
+ def axis_angle_to_quaternion(axis_angle):
450
+ """
451
+ Convert rotations given as axis/angle to quaternions.
452
+
453
+ Args:
454
+ axis_angle: Rotations given as a vector in axis angle form,
455
+ as a tensor of shape (..., 3), where the magnitude is
456
+ the angle turned anticlockwise in radians around the
457
+ vector's direction.
458
+
459
+ Returns:
460
+ quaternions with real part first, as tensor of shape (..., 4).
461
+ """
462
+ angles = torch.norm(axis_angle, p=2, dim=-1, keepdim=True)
463
+ half_angles = 0.5 * angles
464
+ eps = 1e-6
465
+ small_angles = angles.abs() < eps
466
+ sin_half_angles_over_angles = torch.empty_like(angles)
467
+ sin_half_angles_over_angles[~small_angles] = (
468
+ torch.sin(half_angles[~small_angles]) / angles[~small_angles]
469
+ )
470
+ # for x small, sin(x/2) is about x/2 - (x/2)^3/6
471
+ # so sin(x/2)/x is about 1/2 - (x*x)/48
472
+ sin_half_angles_over_angles[small_angles] = (
473
+ 0.5 - (angles[small_angles] * angles[small_angles]) / 48
474
+ )
475
+ quaternions = torch.cat(
476
+ [torch.cos(half_angles), axis_angle * sin_half_angles_over_angles], dim=-1
477
+ )
478
+ return quaternions
479
+
480
+
481
+ def quaternion_to_axis_angle(quaternions):
482
+ """
483
+ Convert rotations given as quaternions to axis/angle.
484
+
485
+ Args:
486
+ quaternions: quaternions with real part first,
487
+ as tensor of shape (..., 4).
488
+
489
+ Returns:
490
+ Rotations given as a vector in axis angle form, as a tensor
491
+ of shape (..., 3), where the magnitude is the angle
492
+ turned anticlockwise in radians around the vector's
493
+ direction.
494
+ """
495
+ norms = torch.norm(quaternions[..., 1:], p=2, dim=-1, keepdim=True)
496
+ half_angles = torch.atan2(norms, quaternions[..., :1])
497
+ angles = 2 * half_angles
498
+ eps = 1e-6
499
+ small_angles = angles.abs() < eps
500
+ sin_half_angles_over_angles = torch.empty_like(angles)
501
+ sin_half_angles_over_angles[~small_angles] = (
502
+ torch.sin(half_angles[~small_angles]) / angles[~small_angles]
503
+ )
504
+ # for x small, sin(x/2) is about x/2 - (x/2)^3/6
505
+ # so sin(x/2)/x is about 1/2 - (x*x)/48
506
+ sin_half_angles_over_angles[small_angles] = (
507
+ 0.5 - (angles[small_angles] * angles[small_angles]) / 48
508
+ )
509
+ return quaternions[..., 1:] / sin_half_angles_over_angles
510
+
511
+
512
+ def rotation_6d_to_matrix(d6: torch.Tensor) -> torch.Tensor:
513
+ """
514
+ Converts 6D rotation representation by Zhou et al. [1] to rotation matrix
515
+ using Gram--Schmidt orthogonalisation per Section B of [1].
516
+ Args:
517
+ d6: 6D rotation representation, of size (*, 6)
518
+
519
+ Returns:
520
+ batch of rotation matrices of size (*, 3, 3)
521
+
522
+ [1] Zhou, Y., Barnes, C., Lu, J., Yang, J., & Li, H.
523
+ On the Continuity of Rotation Representations in Neural Networks.
524
+ IEEE Conference on Computer Vision and Pattern Recognition, 2019.
525
+ Retrieved from http://arxiv.org/abs/1812.07035
526
+ """
527
+
528
+ a1, a2 = d6[..., :3], d6[..., 3:]
529
+ b1 = F.normalize(a1, dim=-1)
530
+ b2 = a2 - (b1 * a2).sum(-1, keepdim=True) * b1
531
+ b2 = F.normalize(b2, dim=-1)
532
+ b3 = torch.cross(b1, b2, dim=-1)
533
+ return torch.stack((b1, b2, b3), dim=-2)
534
+
535
+
536
+ def matrix_to_rotation_6d(matrix: torch.Tensor) -> torch.Tensor:
537
+ """
538
+ Converts rotation matrices to 6D rotation representation by Zhou et al. [1]
539
+ by dropping the last row. Note that 6D representation is not unique.
540
+ Args:
541
+ matrix: batch of rotation matrices of size (*, 3, 3)
542
+
543
+ Returns:
544
+ 6D rotation representation, of size (*, 6)
545
+
546
+ [1] Zhou, Y., Barnes, C., Lu, J., Yang, J., & Li, H.
547
+ On the Continuity of Rotation Representations in Neural Networks.
548
+ IEEE Conference on Computer Vision and Pattern Recognition, 2019.
549
+ Retrieved from http://arxiv.org/abs/1812.07035
550
+ """
551
+ return matrix[..., :2, :].clone().reshape(*matrix.size()[:-2], 6)
data_utils/split_more_than_2s.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2df6e745cdf7473f13ce3ae2ed759c3cceb60c9197e7f3fd65110e7bc20b6f2d
3
+ size 2398875
data_utils/split_train_val_test.py ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import json
3
+ import shutil
4
+
5
+ if __name__ =='__main__':
6
+ id_list = "chemistry conan oliver seth"
7
+ id_list = id_list.split(' ')
8
+
9
+ old_root = '/home/usename/talkshow_data/ExpressiveWholeBodyDatasetReleaseV1.0'
10
+ new_root = '/home/usename/talkshow_data/ExpressiveWholeBodyDatasetReleaseV1.0/talkshow_data_splited'
11
+
12
+ with open('train_val_test.json') as f:
13
+ split_info = json.load(f)
14
+ phase_list = ['train', 'val', 'test']
15
+ for phase in phase_list:
16
+ phase_path_list = split_info[phase]
17
+ for p in phase_path_list:
18
+ old_path = os.path.join(old_root, p)
19
+ if not os.path.exists(old_path):
20
+ print(f'{old_path} not found, continue' )
21
+ continue
22
+ new_path = os.path.join(new_root, phase, p)
23
+ dir_name = os.path.dirname(new_path)
24
+ if not os.path.isdir(dir_name):
25
+ os.makedirs(dir_name, exist_ok=True)
26
+ shutil.move(old_path, new_path)
27
+
data_utils/train_val_test.json ADDED
The diff for this file is too large to render. See raw diff
 
data_utils/utils.py ADDED
@@ -0,0 +1,318 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import numpy as np
2
+ # import librosa #has to do this cause librosa is not supported on my server
3
+ import python_speech_features
4
+ from scipy.io import wavfile
5
+ from scipy import signal
6
+ import librosa
7
+ import torch
8
+ import torchaudio as ta
9
+ import torchaudio.functional as ta_F
10
+ import torchaudio.transforms as ta_T
11
+ # import pyloudnorm as pyln
12
+
13
+
14
+ def load_wav_old(audio_fn, sr = 16000):
15
+ sample_rate, sig = wavfile.read(audio_fn)
16
+ if sample_rate != sr:
17
+ result = int((sig.shape[0]) / sample_rate * sr)
18
+ x_resampled = signal.resample(sig, result)
19
+ x_resampled = x_resampled.astype(np.float64)
20
+ return x_resampled, sr
21
+
22
+ sig = sig / (2**15)
23
+ return sig, sample_rate
24
+
25
+
26
+ def get_mfcc(audio_fn, eps=1e-6, fps=25, smlpx=False, sr=16000, n_mfcc=64, win_size=None):
27
+
28
+ y, sr = librosa.load(audio_fn, sr=sr, mono=True)
29
+
30
+ if win_size is None:
31
+ hop_len=int(sr / fps)
32
+ else:
33
+ hop_len=int(sr / win_size)
34
+
35
+ n_fft=2048
36
+
37
+ C = librosa.feature.mfcc(
38
+ y = y,
39
+ sr = sr,
40
+ n_mfcc = n_mfcc,
41
+ hop_length = hop_len,
42
+ n_fft = n_fft
43
+ )
44
+
45
+ if C.shape[0] == n_mfcc:
46
+ C = C.transpose(1, 0)
47
+
48
+ return C
49
+
50
+
51
+ def get_melspec(audio_fn, eps=1e-6, fps = 25, sr=16000, n_mels=64):
52
+ raise NotImplementedError
53
+ '''
54
+ # y, sr = load_wav(audio_fn=audio_fn, sr=sr)
55
+
56
+ # hop_len = int(sr / fps)
57
+ # n_fft = 2048
58
+
59
+ # C = librosa.feature.melspectrogram(
60
+ # y = y,
61
+ # sr = sr,
62
+ # n_fft=n_fft,
63
+ # hop_length=hop_len,
64
+ # n_mels = n_mels,
65
+ # fmin=0,
66
+ # fmax=8000)
67
+
68
+
69
+ # mask = (C == 0).astype(np.float)
70
+ # C = mask * eps + (1-mask) * C
71
+
72
+ # C = np.log(C)
73
+ # #wierd error may occur here
74
+ # assert not (np.isnan(C).any()), audio_fn
75
+ # if C.shape[0] == n_mels:
76
+ # C = C.transpose(1, 0)
77
+
78
+ # return C
79
+ '''
80
+
81
+ def extract_mfcc(audio,sample_rate=16000):
82
+ mfcc = zip(*python_speech_features.mfcc(audio,sample_rate, numcep=64, nfilt=64, nfft=2048, winstep=0.04))
83
+ mfcc = np.stack([np.array(i) for i in mfcc])
84
+ return mfcc
85
+
86
+ def get_mfcc_psf(audio_fn, eps=1e-6, fps=25, smlpx=False, sr=16000, n_mfcc=64, win_size=None):
87
+ y, sr = load_wav_old(audio_fn, sr=sr)
88
+
89
+ if y.shape.__len__() > 1:
90
+ y = (y[:,0]+y[:,1])/2
91
+
92
+ if win_size is None:
93
+ hop_len=int(sr / fps)
94
+ else:
95
+ hop_len=int(sr/ win_size)
96
+
97
+ n_fft=2048
98
+
99
+ #hard coded for 25 fps
100
+ if not smlpx:
101
+ C = python_speech_features.mfcc(y, sr, numcep=n_mfcc, nfilt=n_mfcc, nfft=n_fft, winstep=0.04)
102
+ else:
103
+ C = python_speech_features.mfcc(y, sr, numcep=n_mfcc, nfilt=n_mfcc, nfft=n_fft, winstep=1.01/15)
104
+ # if C.shape[0] == n_mfcc:
105
+ # C = C.transpose(1, 0)
106
+
107
+ return C
108
+
109
+
110
+ def get_mfcc_psf_min(audio_fn, eps=1e-6, fps=25, smlpx=False, sr=16000, n_mfcc=64, win_size=None):
111
+ y, sr = load_wav_old(audio_fn, sr=sr)
112
+
113
+ if y.shape.__len__() > 1:
114
+ y = (y[:, 0] + y[:, 1]) / 2
115
+ n_fft = 2048
116
+
117
+ slice_len = 22000 * 5
118
+ slice = y.size // slice_len
119
+
120
+ C = []
121
+
122
+ for i in range(slice):
123
+ if i != (slice - 1):
124
+ feat = python_speech_features.mfcc(y[i*slice_len:(i+1)*slice_len], sr, numcep=n_mfcc, nfilt=n_mfcc, nfft=n_fft, winstep=1.01 / 15)
125
+ else:
126
+ feat = python_speech_features.mfcc(y[i * slice_len:], sr, numcep=n_mfcc, nfilt=n_mfcc, nfft=n_fft, winstep=1.01 / 15)
127
+
128
+ C.append(feat)
129
+
130
+ return C
131
+
132
+
133
+ def audio_chunking(audio: torch.Tensor, frame_rate: int = 30, chunk_size: int = 16000):
134
+ """
135
+ :param audio: 1 x T tensor containing a 16kHz audio signal
136
+ :param frame_rate: frame rate for video (we need one audio chunk per video frame)
137
+ :param chunk_size: number of audio samples per chunk
138
+ :return: num_chunks x chunk_size tensor containing sliced audio
139
+ """
140
+ samples_per_frame = chunk_size // frame_rate
141
+ padding = (chunk_size - samples_per_frame) // 2
142
+ audio = torch.nn.functional.pad(audio.unsqueeze(0), pad=[padding, padding]).squeeze(0)
143
+ anchor_points = list(range(chunk_size//2, audio.shape[-1]-chunk_size//2, samples_per_frame))
144
+ audio = torch.cat([audio[:, i-chunk_size//2:i+chunk_size//2] for i in anchor_points], dim=0)
145
+ return audio
146
+
147
+
148
+ def get_mfcc_ta(audio_fn, eps=1e-6, fps=15, smlpx=False, sr=16000, n_mfcc=64, win_size=None, type='mfcc', am=None, am_sr=None, encoder_choice='mfcc'):
149
+ if am is None:
150
+ audio, sr_0 = ta.load(audio_fn)
151
+ if sr != sr_0:
152
+ audio = ta.transforms.Resample(sr_0, sr)(audio)
153
+ if audio.shape[0] > 1:
154
+ audio = torch.mean(audio, dim=0, keepdim=True)
155
+
156
+ n_fft = 2048
157
+ if fps == 15:
158
+ hop_length = 1467
159
+ elif fps == 30:
160
+ hop_length = 734
161
+ win_length = hop_length * 2
162
+ n_mels = 256
163
+ n_mfcc = 64
164
+
165
+ if type == 'mfcc':
166
+ mfcc_transform = ta_T.MFCC(
167
+ sample_rate=sr,
168
+ n_mfcc=n_mfcc,
169
+ melkwargs={
170
+ "n_fft": n_fft,
171
+ "n_mels": n_mels,
172
+ # "win_length": win_length,
173
+ "hop_length": hop_length,
174
+ "mel_scale": "htk",
175
+ },
176
+ )
177
+ audio_ft = mfcc_transform(audio).squeeze(dim=0).transpose(0,1).numpy()
178
+ elif type == 'mel':
179
+ # audio = 0.01 * audio / torch.mean(torch.abs(audio))
180
+ mel_transform = ta_T.MelSpectrogram(
181
+ sample_rate=sr, n_fft=n_fft, win_length=None, hop_length=hop_length, n_mels=n_mels
182
+ )
183
+ audio_ft = mel_transform(audio).squeeze(0).transpose(0,1).numpy()
184
+ # audio_ft = torch.log(audio_ft.clamp(min=1e-10, max=None)).transpose(0,1).numpy()
185
+ elif type == 'mel_mul':
186
+ audio = 0.01 * audio / torch.mean(torch.abs(audio))
187
+ audio = audio_chunking(audio, frame_rate=fps, chunk_size=sr)
188
+ mel_transform = ta_T.MelSpectrogram(
189
+ sample_rate=sr, n_fft=n_fft, win_length=int(sr/20), hop_length=int(sr/100), n_mels=n_mels
190
+ )
191
+ audio_ft = mel_transform(audio).squeeze(1)
192
+ audio_ft = torch.log(audio_ft.clamp(min=1e-10, max=None)).numpy()
193
+ else:
194
+ speech_array, sampling_rate = librosa.load(audio_fn, sr=16000)
195
+
196
+ if encoder_choice == 'faceformer':
197
+ # audio_ft = np.squeeze(am(speech_array, sampling_rate=16000).input_values).reshape(-1, 1)
198
+ audio_ft = speech_array.reshape(-1, 1)
199
+ elif encoder_choice == 'meshtalk':
200
+ audio_ft = 0.01 * speech_array / np.mean(np.abs(speech_array))
201
+ elif encoder_choice == 'onset':
202
+ audio_ft = librosa.onset.onset_detect(y=speech_array, sr=16000, units='time').reshape(-1, 1)
203
+ else:
204
+ audio, sr_0 = ta.load(audio_fn)
205
+ if sr != sr_0:
206
+ audio = ta.transforms.Resample(sr_0, sr)(audio)
207
+ if audio.shape[0] > 1:
208
+ audio = torch.mean(audio, dim=0, keepdim=True)
209
+
210
+ n_fft = 2048
211
+ if fps == 15:
212
+ hop_length = 1467
213
+ elif fps == 30:
214
+ hop_length = 734
215
+ win_length = hop_length * 2
216
+ n_mels = 256
217
+ n_mfcc = 64
218
+
219
+ mfcc_transform = ta_T.MFCC(
220
+ sample_rate=sr,
221
+ n_mfcc=n_mfcc,
222
+ melkwargs={
223
+ "n_fft": n_fft,
224
+ "n_mels": n_mels,
225
+ # "win_length": win_length,
226
+ "hop_length": hop_length,
227
+ "mel_scale": "htk",
228
+ },
229
+ )
230
+ audio_ft = mfcc_transform(audio).squeeze(dim=0).transpose(0, 1).numpy()
231
+ return audio_ft
232
+
233
+
234
+ def get_mfcc_sepa(audio_fn, fps=15, sr=16000):
235
+ audio, sr_0 = ta.load(audio_fn)
236
+ if sr != sr_0:
237
+ audio = ta.transforms.Resample(sr_0, sr)(audio)
238
+ if audio.shape[0] > 1:
239
+ audio = torch.mean(audio, dim=0, keepdim=True)
240
+
241
+ n_fft = 2048
242
+ if fps == 15:
243
+ hop_length = 1467
244
+ elif fps == 30:
245
+ hop_length = 734
246
+ n_mels = 256
247
+ n_mfcc = 64
248
+
249
+ mfcc_transform = ta_T.MFCC(
250
+ sample_rate=sr,
251
+ n_mfcc=n_mfcc,
252
+ melkwargs={
253
+ "n_fft": n_fft,
254
+ "n_mels": n_mels,
255
+ # "win_length": win_length,
256
+ "hop_length": hop_length,
257
+ "mel_scale": "htk",
258
+ },
259
+ )
260
+ audio_ft_0 = mfcc_transform(audio[0, :sr*2]).squeeze(dim=0).transpose(0,1).numpy()
261
+ audio_ft_1 = mfcc_transform(audio[0, sr*2:]).squeeze(dim=0).transpose(0,1).numpy()
262
+ audio_ft = np.concatenate((audio_ft_0, audio_ft_1), axis=0)
263
+ return audio_ft, audio_ft_0.shape[0]
264
+
265
+
266
+ def get_mfcc_old(wav_file):
267
+ sig, sample_rate = load_wav_old(wav_file)
268
+ mfcc = extract_mfcc(sig)
269
+ return mfcc
270
+
271
+
272
+ def smooth_geom(geom, mask: torch.Tensor = None, filter_size: int = 9, sigma: float = 2.0):
273
+ """
274
+ :param geom: T x V x 3 tensor containing a temporal sequence of length T with V vertices in each frame
275
+ :param mask: V-dimensional Tensor containing a mask with vertices to be smoothed
276
+ :param filter_size: size of the Gaussian filter
277
+ :param sigma: standard deviation of the Gaussian filter
278
+ :return: T x V x 3 tensor containing smoothed geometry (i.e., smoothed in the area indicated by the mask)
279
+ """
280
+ assert filter_size % 2 == 1, f"filter size must be odd but is {filter_size}"
281
+ # Gaussian smoothing (low-pass filtering)
282
+ fltr = np.arange(-(filter_size // 2), filter_size // 2 + 1)
283
+ fltr = np.exp(-0.5 * fltr ** 2 / sigma ** 2)
284
+ fltr = torch.Tensor(fltr) / np.sum(fltr)
285
+ # apply fltr
286
+ fltr = fltr.view(1, 1, -1).to(device=geom.device)
287
+ T, V = geom.shape[1], geom.shape[2]
288
+ g = torch.nn.functional.pad(
289
+ geom.permute(2, 0, 1).view(V, 1, T),
290
+ pad=[filter_size // 2, filter_size // 2], mode='replicate'
291
+ )
292
+ g = torch.nn.functional.conv1d(g, fltr).view(V, 1, T)
293
+ smoothed = g.permute(1, 2, 0).contiguous()
294
+ # blend smoothed signal with original signal
295
+ if mask is None:
296
+ return smoothed
297
+ else:
298
+ return smoothed * mask[None, :, None] + geom * (-mask[None, :, None] + 1)
299
+
300
+ if __name__ == '__main__':
301
+ audio_fn = '../sample_audio/clip000028_tCAkv4ggPgI.wav'
302
+
303
+ C = get_mfcc_psf(audio_fn)
304
+ print(C.shape)
305
+
306
+ C_2 = get_mfcc_librosa(audio_fn)
307
+ print(C.shape)
308
+
309
+ print(C)
310
+ print(C_2)
311
+ print((C == C_2).all())
312
+ # print(y.shape, sr)
313
+ # mel_spec = get_melspec(audio_fn)
314
+ # print(mel_spec.shape)
315
+ # mfcc = get_mfcc(audio_fn, sr = 16000)
316
+ # print(mfcc.shape)
317
+ # print(mel_spec.max(), mel_spec.min())
318
+ # print(mfcc.max(), mfcc.min())
evaluation/FGD.py ADDED
@@ -0,0 +1,199 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import time
2
+
3
+ import numpy as np
4
+ import torch
5
+ import torch.nn.functional as F
6
+ from scipy import linalg
7
+ import math
8
+ from data_utils.rotation_conversion import axis_angle_to_matrix, matrix_to_rotation_6d
9
+
10
+ import warnings
11
+ warnings.filterwarnings("ignore", category=RuntimeWarning) # ignore warnings
12
+
13
+
14
+ change_angle = torch.tensor([6.0181e-05, 5.1597e-05, 2.1344e-04, 2.1899e-04])
15
+ class EmbeddingSpaceEvaluator:
16
+ def __init__(self, ae, vae, device):
17
+
18
+ # init embed net
19
+ self.ae = ae
20
+ # self.vae = vae
21
+
22
+ # storage
23
+ self.real_feat_list = []
24
+ self.generated_feat_list = []
25
+ self.real_joints_list = []
26
+ self.generated_joints_list = []
27
+ self.real_6d_list = []
28
+ self.generated_6d_list = []
29
+ self.audio_beat_list = []
30
+
31
+ def reset(self):
32
+ self.real_feat_list = []
33
+ self.generated_feat_list = []
34
+
35
+ def get_no_of_samples(self):
36
+ return len(self.real_feat_list)
37
+
38
+ def push_samples(self, generated_poses, real_poses):
39
+ # self.net.eval()
40
+ # convert poses to latent features
41
+ real_feat, real_poses = self.ae.extract(real_poses)
42
+ generated_feat, generated_poses = self.ae.extract(generated_poses)
43
+
44
+ num_joints = real_poses.shape[2] // 3
45
+
46
+ real_feat = real_feat.squeeze()
47
+ generated_feat = generated_feat.reshape(generated_feat.shape[0]*generated_feat.shape[1], -1)
48
+
49
+ self.real_feat_list.append(real_feat.data.cpu().numpy())
50
+ self.generated_feat_list.append(generated_feat.data.cpu().numpy())
51
+
52
+ # real_poses = matrix_to_rotation_6d(axis_angle_to_matrix(real_poses.reshape(-1, 3))).reshape(-1, num_joints, 6)
53
+ # generated_poses = matrix_to_rotation_6d(axis_angle_to_matrix(generated_poses.reshape(-1, 3))).reshape(-1, num_joints, 6)
54
+ #
55
+ # self.real_feat_list.append(real_poses.data.cpu().numpy())
56
+ # self.generated_feat_list.append(generated_poses.data.cpu().numpy())
57
+
58
+ def push_joints(self, generated_poses, real_poses):
59
+ self.real_joints_list.append(real_poses.data.cpu())
60
+ self.generated_joints_list.append(generated_poses.squeeze().data.cpu())
61
+
62
+ def push_aud(self, aud):
63
+ self.audio_beat_list.append(aud.squeeze().data.cpu())
64
+
65
+ def get_MAAC(self):
66
+ ang_vel_list = []
67
+ for real_joints in self.real_joints_list:
68
+ real_joints[:, 15:21] = real_joints[:, 16:22]
69
+ vec = real_joints[:, 15:21] - real_joints[:, 13:19]
70
+ inner_product = torch.einsum('kij,kij->ki', [vec[:, 2:], vec[:, :-2]])
71
+ inner_product = torch.clamp(inner_product, -1, 1, out=None)
72
+ angle = torch.acos(inner_product) / math.pi
73
+ ang_vel = (angle[1:] - angle[:-1]).abs().mean(dim=0)
74
+ ang_vel_list.append(ang_vel.unsqueeze(dim=0))
75
+ all_vel = torch.cat(ang_vel_list, dim=0)
76
+ MAAC = all_vel.mean(dim=0)
77
+ return MAAC
78
+
79
+ def get_BCscore(self):
80
+ thres = 0.01
81
+ sigma = 0.1
82
+ sum_1 = 0
83
+ total_beat = 0
84
+ for joints, audio_beat_time in zip(self.generated_joints_list, self.audio_beat_list):
85
+ motion_beat_time = []
86
+ if joints.dim() == 4:
87
+ joints = joints[0]
88
+ joints[:, 15:21] = joints[:, 16:22]
89
+ vec = joints[:, 15:21] - joints[:, 13:19]
90
+ inner_product = torch.einsum('kij,kij->ki', [vec[:, 2:], vec[:, :-2]])
91
+ inner_product = torch.clamp(inner_product, -1, 1, out=None)
92
+ angle = torch.acos(inner_product) / math.pi
93
+ ang_vel = (angle[1:] - angle[:-1]).abs() / change_angle / len(change_angle)
94
+
95
+ angle_diff = torch.cat((torch.zeros(1, 4), ang_vel), dim=0)
96
+
97
+ sum_2 = 0
98
+ for i in range(angle_diff.shape[1]):
99
+ motion_beat_time = []
100
+ for t in range(1, joints.shape[0]-1):
101
+ if (angle_diff[t][i] < angle_diff[t - 1][i] and angle_diff[t][i] < angle_diff[t + 1][i]):
102
+ if (angle_diff[t - 1][i] - angle_diff[t][i] >= thres or angle_diff[t + 1][i] - angle_diff[
103
+ t][i] >= thres):
104
+ motion_beat_time.append(float(t) / 30.0)
105
+ if (len(motion_beat_time) == 0):
106
+ continue
107
+ motion_beat_time = torch.tensor(motion_beat_time)
108
+ sum = 0
109
+ for audio in audio_beat_time:
110
+ sum += np.power(math.e, -(np.power((audio.item() - motion_beat_time), 2)).min() / (2 * sigma * sigma))
111
+ sum_2 = sum_2 + sum
112
+ total_beat = total_beat + len(audio_beat_time)
113
+ sum_1 = sum_1 + sum_2
114
+ return sum_1/total_beat
115
+
116
+
117
+ def get_scores(self):
118
+ generated_feats = np.vstack(self.generated_feat_list)
119
+ real_feats = np.vstack(self.real_feat_list)
120
+
121
+ def frechet_distance(samples_A, samples_B):
122
+ A_mu = np.mean(samples_A, axis=0)
123
+ A_sigma = np.cov(samples_A, rowvar=False)
124
+ B_mu = np.mean(samples_B, axis=0)
125
+ B_sigma = np.cov(samples_B, rowvar=False)
126
+ try:
127
+ frechet_dist = self.calculate_frechet_distance(A_mu, A_sigma, B_mu, B_sigma)
128
+ except ValueError:
129
+ frechet_dist = 1e+10
130
+ return frechet_dist
131
+
132
+ ####################################################################
133
+ # frechet distance
134
+ frechet_dist = frechet_distance(generated_feats, real_feats)
135
+
136
+ ####################################################################
137
+ # distance between real and generated samples on the latent feature space
138
+ dists = []
139
+ for i in range(real_feats.shape[0]):
140
+ d = np.sum(np.absolute(real_feats[i] - generated_feats[i])) # MAE
141
+ dists.append(d)
142
+ feat_dist = np.mean(dists)
143
+
144
+ return frechet_dist, feat_dist
145
+
146
+ @staticmethod
147
+ def calculate_frechet_distance(mu1, sigma1, mu2, sigma2, eps=1e-6):
148
+ """ from https://github.com/mseitzer/pytorch-fid/blob/master/fid_score.py """
149
+ """Numpy implementation of the Frechet Distance.
150
+ The Frechet distance between two multivariate Gaussians X_1 ~ N(mu_1, C_1)
151
+ and X_2 ~ N(mu_2, C_2) is
152
+ d^2 = ||mu_1 - mu_2||^2 + Tr(C_1 + C_2 - 2*sqrt(C_1*C_2)).
153
+ Stable version by Dougal J. Sutherland.
154
+ Params:
155
+ -- mu1 : Numpy array containing the activations of a layer of the
156
+ inception net (like returned by the function 'get_predictions')
157
+ for generated samples.
158
+ -- mu2 : The sample mean over activations, precalculated on an
159
+ representative data set.
160
+ -- sigma1: The covariance matrix over activations for generated samples.
161
+ -- sigma2: The covariance matrix over activations, precalculated on an
162
+ representative data set.
163
+ Returns:
164
+ -- : The Frechet Distance.
165
+ """
166
+
167
+ mu1 = np.atleast_1d(mu1)
168
+ mu2 = np.atleast_1d(mu2)
169
+
170
+ sigma1 = np.atleast_2d(sigma1)
171
+ sigma2 = np.atleast_2d(sigma2)
172
+
173
+ assert mu1.shape == mu2.shape, \
174
+ 'Training and test mean vectors have different lengths'
175
+ assert sigma1.shape == sigma2.shape, \
176
+ 'Training and test covariances have different dimensions'
177
+
178
+ diff = mu1 - mu2
179
+
180
+ # Product might be almost singular
181
+ covmean, _ = linalg.sqrtm(sigma1.dot(sigma2), disp=False)
182
+ if not np.isfinite(covmean).all():
183
+ msg = ('fid calculation produces singular product; '
184
+ 'adding %s to diagonal of cov estimates') % eps
185
+ print(msg)
186
+ offset = np.eye(sigma1.shape[0]) * eps
187
+ covmean = linalg.sqrtm((sigma1 + offset).dot(sigma2 + offset))
188
+
189
+ # Numerical error might give slight imaginary component
190
+ if np.iscomplexobj(covmean):
191
+ if not np.allclose(np.diagonal(covmean).imag, 0, atol=1e-3):
192
+ m = np.max(np.abs(covmean.imag))
193
+ raise ValueError('Imaginary component {}'.format(m))
194
+ covmean = covmean.real
195
+
196
+ tr_covmean = np.trace(covmean)
197
+
198
+ return (diff.dot(diff) + np.trace(sigma1) +
199
+ np.trace(sigma2) - 2 * tr_covmean)
evaluation/__init__.py ADDED
File without changes
evaluation/__pycache__/__init__.cpython-37.pyc ADDED
Binary file (181 Bytes). View file
 
evaluation/__pycache__/metrics.cpython-37.pyc ADDED
Binary file (3.81 kB). View file
 
evaluation/diversity_LVD.py ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ '''
2
+ LVD: different initial pose
3
+ diversity: same initial pose
4
+ '''
5
+ import os
6
+ import sys
7
+ sys.path.append(os.getcwd())
8
+
9
+ from glob import glob
10
+
11
+ from argparse import ArgumentParser
12
+ import json
13
+
14
+ from evaluation.util import *
15
+ from evaluation.metrics import *
16
+ from tqdm import tqdm
17
+
18
+ parser = ArgumentParser()
19
+ parser.add_argument('--speaker', required=True, type=str)
20
+ parser.add_argument('--post_fix', nargs='+', default=['base'], type=str)
21
+ args = parser.parse_args()
22
+
23
+ speaker = args.speaker
24
+ test_audios = sorted(glob('pose_dataset/videos/test_audios/%s/*.wav'%(speaker)))
25
+
26
+ LVD_list = []
27
+ diversity_list = []
28
+
29
+ for aud in tqdm(test_audios):
30
+ base_name = os.path.splitext(aud)[0]
31
+ gt_path = get_full_path(aud, speaker, 'val')
32
+ _, gt_poses, _ = get_gts(gt_path)
33
+ gt_poses = gt_poses[np.newaxis,...]
34
+ # print(gt_poses.shape)#(seq_len, 135*2)pose, lhand, rhand, face
35
+ for post_fix in args.post_fix:
36
+ pred_path = base_name + '_'+post_fix+'.json'
37
+ pred_poses = np.array(json.load(open(pred_path)))
38
+ # print(pred_poses.shape)#(B, seq_len, 108)
39
+ pred_poses = cvt25(pred_poses, gt_poses)
40
+ # print(pred_poses.shape)#(B, seq, pose_dim)
41
+
42
+ gt_valid_points = hand_points(gt_poses)
43
+ pred_valid_points = hand_points(pred_poses)
44
+
45
+ lvd = LVD(gt_valid_points, pred_valid_points)
46
+ # div = diversity(pred_valid_points)
47
+
48
+ LVD_list.append(lvd)
49
+ # diversity_list.append(div)
50
+
51
+ # gt_velocity = peak_velocity(gt_valid_points, order=2)
52
+ # pred_velocity = peak_velocity(pred_valid_points, order=2)
53
+
54
+ # gt_consistency = velocity_consistency(gt_velocity, pred_velocity)
55
+ # pred_consistency = velocity_consistency(pred_velocity, gt_velocity)
56
+
57
+ # gt_consistency_list.append(gt_consistency)
58
+ # pred_consistency_list.append(pred_consistency)
59
+
60
+ lvd = np.mean(LVD_list)
61
+ # diversity_list = np.mean(diversity_list)
62
+
63
+ print('LVD:', lvd)
64
+ # print("diversity:", diversity_list)
evaluation/get_quality_samples.py ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ '''
2
+ '''
3
+ import os
4
+ import sys
5
+ sys.path.append(os.getcwd())
6
+
7
+ from glob import glob
8
+
9
+ from argparse import ArgumentParser
10
+ import json
11
+
12
+ from evaluation.util import *
13
+ from evaluation.metrics import *
14
+ from tqdm import tqdm
15
+
16
+ parser = ArgumentParser()
17
+ parser.add_argument('--speaker', required=True, type=str)
18
+ parser.add_argument('--post_fix', nargs='+', default=['paper_model'], type=str)
19
+ args = parser.parse_args()
20
+
21
+ speaker = args.speaker
22
+ test_audios = sorted(glob('pose_dataset/videos/test_audios/%s/*.wav'%(speaker)))
23
+
24
+ quality_samples={'gt':[]}
25
+ for post_fix in args.post_fix:
26
+ quality_samples[post_fix] = []
27
+
28
+ for aud in tqdm(test_audios):
29
+ base_name = os.path.splitext(aud)[0]
30
+ gt_path = get_full_path(aud, speaker, 'val')
31
+ _, gt_poses, _ = get_gts(gt_path)
32
+ gt_poses = gt_poses[np.newaxis,...]
33
+ gt_valid_points = valid_points(gt_poses)
34
+ # print(gt_valid_points.shape)
35
+ quality_samples['gt'].append(gt_valid_points)
36
+
37
+ for post_fix in args.post_fix:
38
+ pred_path = base_name + '_'+post_fix+'.json'
39
+ pred_poses = np.array(json.load(open(pred_path)))
40
+ # print(pred_poses.shape)#(B, seq_len, 108)
41
+ pred_poses = cvt25(pred_poses, gt_poses)
42
+ # print(pred_poses.shape)#(B, seq, pose_dim)
43
+
44
+ pred_valid_points = valid_points(pred_poses)[0:1]
45
+ quality_samples[post_fix].append(pred_valid_points)
46
+
47
+ quality_samples['gt'] = np.concatenate(quality_samples['gt'], axis=1)
48
+ for post_fix in args.post_fix:
49
+ quality_samples[post_fix] = np.concatenate(quality_samples[post_fix], axis=1)
50
+
51
+ print('gt:', quality_samples['gt'].shape)
52
+ quality_samples['gt'] = quality_samples['gt'].tolist()
53
+ for post_fix in args.post_fix:
54
+ print(post_fix, ':', quality_samples[post_fix].shape)
55
+ quality_samples[post_fix] = quality_samples[post_fix].tolist()
56
+
57
+ save_dir = '../../experiments/'
58
+ os.makedirs(save_dir, exist_ok=True)
59
+ save_name = os.path.join(save_dir, 'quality_samples_%s.json'%(speaker))
60
+ with open(save_name, 'w') as f:
61
+ json.dump(quality_samples, f)
62
+
evaluation/metrics.py ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ '''
2
+ Warning: metrics are for reference only, may have limited significance
3
+ '''
4
+ import os
5
+ import sys
6
+ sys.path.append(os.getcwd())
7
+ import numpy as np
8
+ import torch
9
+
10
+ from data_utils.lower_body import rearrange, symmetry
11
+ import torch.nn.functional as F
12
+
13
+ def data_driven_baselines(gt_kps):
14
+ '''
15
+ gt_kps: T, D
16
+ '''
17
+ gt_velocity = np.abs(gt_kps[1:] - gt_kps[:-1])
18
+
19
+ mean= np.mean(gt_velocity, axis=0)[np.newaxis] #(1, D)
20
+ mean = np.mean(np.abs(gt_velocity-mean))
21
+ last_step = gt_kps[1] - gt_kps[0]
22
+ last_step = last_step[np.newaxis] #(1, D)
23
+ last_step = np.mean(np.abs(gt_velocity-last_step))
24
+ return last_step, mean
25
+
26
+ def Batch_LVD(gt_kps, pr_kps, symmetrical, weight):
27
+ if gt_kps.shape[0] > pr_kps.shape[1]:
28
+ length = pr_kps.shape[1]
29
+ else:
30
+ length = gt_kps.shape[0]
31
+ gt_kps = gt_kps[:length]
32
+ pr_kps = pr_kps[:, :length]
33
+ global symmetry
34
+ symmetry = torch.tensor(symmetry).bool()
35
+
36
+ if symmetrical:
37
+ # rearrange for compute symmetric. ns means non-symmetrical joints, ys means symmetrical joints.
38
+ gt_kps = gt_kps[:, rearrange]
39
+ ns_gt_kps = gt_kps[:, ~symmetry]
40
+ ys_gt_kps = gt_kps[:, symmetry]
41
+ ys_gt_kps = ys_gt_kps.reshape(ys_gt_kps.shape[0], -1, 2, 3)
42
+ ns_gt_velocity = (ns_gt_kps[1:] - ns_gt_kps[:-1]).norm(p=2, dim=-1)
43
+ ys_gt_velocity = (ys_gt_kps[1:] - ys_gt_kps[:-1]).norm(p=2, dim=-1)
44
+ left_gt_vel = ys_gt_velocity[:, :, 0].sum(dim=-1)
45
+ right_gt_vel = ys_gt_velocity[:, :, 1].sum(dim=-1)
46
+ move_side = torch.where(left_gt_vel>right_gt_vel, torch.ones(left_gt_vel.shape).cuda(), torch.zeros(left_gt_vel.shape).cuda())
47
+ ys_gt_velocity = torch.mul(ys_gt_velocity[:, :, 0].transpose(0,1), move_side) + torch.mul(ys_gt_velocity[:, :, 1].transpose(0,1), ~move_side.bool())
48
+ ys_gt_velocity = ys_gt_velocity.transpose(0,1)
49
+ gt_velocity = torch.cat([ns_gt_velocity, ys_gt_velocity], dim=1)
50
+
51
+ pr_kps = pr_kps[:, :, rearrange]
52
+ ns_pr_kps = pr_kps[:, :, ~symmetry]
53
+ ys_pr_kps = pr_kps[:, :, symmetry]
54
+ ys_pr_kps = ys_pr_kps.reshape(ys_pr_kps.shape[0], ys_pr_kps.shape[1], -1, 2, 3)
55
+ ns_pr_velocity = (ns_pr_kps[:, 1:] - ns_pr_kps[:, :-1]).norm(p=2, dim=-1)
56
+ ys_pr_velocity = (ys_pr_kps[:, 1:] - ys_pr_kps[:, :-1]).norm(p=2, dim=-1)
57
+ left_pr_vel = ys_pr_velocity[:, :, :, 0].sum(dim=-1)
58
+ right_pr_vel = ys_pr_velocity[:, :, :, 1].sum(dim=-1)
59
+ move_side = torch.where(left_pr_vel > right_pr_vel, torch.ones(left_pr_vel.shape).cuda(),
60
+ torch.zeros(left_pr_vel.shape).cuda())
61
+ ys_pr_velocity = torch.mul(ys_pr_velocity[..., 0].permute(2, 0, 1), move_side) + torch.mul(
62
+ ys_pr_velocity[..., 1].permute(2, 0, 1), ~move_side.long())
63
+ ys_pr_velocity = ys_pr_velocity.permute(1, 2, 0)
64
+ pr_velocity = torch.cat([ns_pr_velocity, ys_pr_velocity], dim=2)
65
+ else:
66
+ gt_velocity = (gt_kps[1:] - gt_kps[:-1]).norm(p=2, dim=-1)
67
+ pr_velocity = (pr_kps[:, 1:] - pr_kps[:, :-1]).norm(p=2, dim=-1)
68
+
69
+ if weight:
70
+ w = F.softmax(gt_velocity.sum(dim=1).normal_(), dim=0)
71
+ else:
72
+ w = 1 / gt_velocity.shape[0]
73
+
74
+ v_diff = ((pr_velocity - gt_velocity).abs().sum(dim=-1) * w).sum(dim=-1).mean()
75
+
76
+ return v_diff
77
+
78
+
79
+ def LVD(gt_kps, pr_kps, symmetrical=False, weight=False):
80
+ gt_kps = gt_kps.squeeze()
81
+ pr_kps = pr_kps.squeeze()
82
+ if len(pr_kps.shape) == 4:
83
+ return Batch_LVD(gt_kps, pr_kps, symmetrical, weight)
84
+ # length = np.minimum(gt_kps.shape[0], pr_kps.shape[0])
85
+ length = gt_kps.shape[0]-10
86
+ # gt_kps = gt_kps[25:length]
87
+ # pr_kps = pr_kps[25:length] #(T, D)
88
+ # if pr_kps.shape[0] < gt_kps.shape[0]:
89
+ # pr_kps = np.pad(pr_kps, [[0, int(gt_kps.shape[0]-pr_kps.shape[0])], [0, 0]], mode='constant')
90
+
91
+ gt_velocity = (gt_kps[1:] - gt_kps[:-1]).norm(p=2, dim=-1)
92
+ pr_velocity = (pr_kps[1:] - pr_kps[:-1]).norm(p=2, dim=-1)
93
+
94
+ return (pr_velocity-gt_velocity).abs().sum(dim=-1).mean()
95
+
96
+ def diversity(kps):
97
+ '''
98
+ kps: bs, seq, dim
99
+ '''
100
+ dis_list = []
101
+ #the distance between each pair
102
+ for i in range(kps.shape[0]):
103
+ for j in range(i+1, kps.shape[0]):
104
+ seq_i = kps[i]
105
+ seq_j = kps[j]
106
+
107
+ dis = np.mean(np.abs(seq_i - seq_j))
108
+ dis_list.append(dis)
109
+ return np.mean(dis_list)
evaluation/mode_transition.py ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import sys
3
+ sys.path.append(os.getcwd())
4
+
5
+ from glob import glob
6
+
7
+ from argparse import ArgumentParser
8
+ import json
9
+
10
+ from evaluation.util import *
11
+ from evaluation.metrics import *
12
+ from tqdm import tqdm
13
+
14
+ parser = ArgumentParser()
15
+ parser.add_argument('--speaker', required=True, type=str)
16
+ parser.add_argument('--post_fix', nargs='+', default=['paper_model'], type=str)
17
+ args = parser.parse_args()
18
+
19
+ speaker = args.speaker
20
+ test_audios = sorted(glob('pose_dataset/videos/test_audios/%s/*.wav'%(speaker)))
21
+
22
+ precision_list=[]
23
+ recall_list=[]
24
+ accuracy_list=[]
25
+
26
+ for aud in tqdm(test_audios):
27
+ base_name = os.path.splitext(aud)[0]
28
+ gt_path = get_full_path(aud, speaker, 'val')
29
+ _, gt_poses, _ = get_gts(gt_path)
30
+ if gt_poses.shape[0] < 50:
31
+ continue
32
+ gt_poses = gt_poses[np.newaxis,...]
33
+ # print(gt_poses.shape)#(seq_len, 135*2)pose, lhand, rhand, face
34
+ for post_fix in args.post_fix:
35
+ pred_path = base_name + '_'+post_fix+'.json'
36
+ pred_poses = np.array(json.load(open(pred_path)))
37
+ # print(pred_poses.shape)#(B, seq_len, 108)
38
+ pred_poses = cvt25(pred_poses, gt_poses)
39
+ # print(pred_poses.shape)#(B, seq, pose_dim)
40
+
41
+ gt_valid_points = valid_points(gt_poses)
42
+ pred_valid_points = valid_points(pred_poses)
43
+
44
+ # print(gt_valid_points.shape, pred_valid_points.shape)
45
+
46
+ gt_mode_transition_seq = mode_transition_seq(gt_valid_points, speaker)#(B, N)
47
+ pred_mode_transition_seq = mode_transition_seq(pred_valid_points, speaker)#(B, N)
48
+
49
+ # baseline = np.random.randint(0, 2, size=pred_mode_transition_seq.shape)
50
+ # pred_mode_transition_seq = baseline
51
+ precision, recall, accuracy = mode_transition_consistency(pred_mode_transition_seq, gt_mode_transition_seq)
52
+ precision_list.append(precision)
53
+ recall_list.append(recall)
54
+ accuracy_list.append(accuracy)
55
+ print(len(precision_list), len(recall_list), len(accuracy_list))
56
+ precision_list = np.mean(precision_list)
57
+ recall_list = np.mean(recall_list)
58
+ accuracy_list = np.mean(accuracy_list)
59
+
60
+ print('precision, recall, accu:', precision_list, recall_list, accuracy_list)
evaluation/peak_velocity.py ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import sys
3
+ sys.path.append(os.getcwd())
4
+
5
+ from glob import glob
6
+
7
+ from argparse import ArgumentParser
8
+ import json
9
+
10
+ from evaluation.util import *
11
+ from evaluation.metrics import *
12
+ from tqdm import tqdm
13
+
14
+ parser = ArgumentParser()
15
+ parser.add_argument('--speaker', required=True, type=str)
16
+ parser.add_argument('--post_fix', nargs='+', default=['paper_model'], type=str)
17
+ args = parser.parse_args()
18
+
19
+ speaker = args.speaker
20
+ test_audios = sorted(glob('pose_dataset/videos/test_audios/%s/*.wav'%(speaker)))
21
+
22
+ gt_consistency_list=[]
23
+ pred_consistency_list=[]
24
+
25
+ for aud in tqdm(test_audios):
26
+ base_name = os.path.splitext(aud)[0]
27
+ gt_path = get_full_path(aud, speaker, 'val')
28
+ _, gt_poses, _ = get_gts(gt_path)
29
+ gt_poses = gt_poses[np.newaxis,...]
30
+ # print(gt_poses.shape)#(seq_len, 135*2)pose, lhand, rhand, face
31
+ for post_fix in args.post_fix:
32
+ pred_path = base_name + '_'+post_fix+'.json'
33
+ pred_poses = np.array(json.load(open(pred_path)))
34
+ # print(pred_poses.shape)#(B, seq_len, 108)
35
+ pred_poses = cvt25(pred_poses, gt_poses)
36
+ # print(pred_poses.shape)#(B, seq, pose_dim)
37
+
38
+ gt_valid_points = hand_points(gt_poses)
39
+ pred_valid_points = hand_points(pred_poses)
40
+
41
+ gt_velocity = peak_velocity(gt_valid_points, order=2)
42
+ pred_velocity = peak_velocity(pred_valid_points, order=2)
43
+
44
+ gt_consistency = velocity_consistency(gt_velocity, pred_velocity)
45
+ pred_consistency = velocity_consistency(pred_velocity, gt_velocity)
46
+
47
+ gt_consistency_list.append(gt_consistency)
48
+ pred_consistency_list.append(pred_consistency)
49
+
50
+ gt_consistency_list = np.concatenate(gt_consistency_list)
51
+ pred_consistency_list = np.concatenate(pred_consistency_list)
52
+
53
+ print(gt_consistency_list.max(), gt_consistency_list.min())
54
+ print(pred_consistency_list.max(), pred_consistency_list.min())
55
+ print(np.mean(gt_consistency_list), np.mean(pred_consistency_list))
56
+ print(np.std(gt_consistency_list), np.std(pred_consistency_list))
57
+
58
+ draw_cdf(gt_consistency_list, save_name='%s_gt.jpg'%(speaker), color='slateblue')
59
+ draw_cdf(pred_consistency_list, save_name='%s_pred.jpg'%(speaker), color='lightskyblue')
60
+
61
+ to_excel(gt_consistency_list, '%s_gt.xlsx'%(speaker))
62
+ to_excel(pred_consistency_list, '%s_pred.xlsx'%(speaker))
63
+
64
+ np.save('%s_gt.npy'%(speaker), gt_consistency_list)
65
+ np.save('%s_pred.npy'%(speaker), pred_consistency_list)
evaluation/util.py ADDED
@@ -0,0 +1,148 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ from glob import glob
3
+ import numpy as np
4
+ import json
5
+ from matplotlib import pyplot as plt
6
+ import pandas as pd
7
+ def get_gts(clip):
8
+ '''
9
+ clip: abs path to the clip dir
10
+ '''
11
+ keypoints_files = sorted(glob(os.path.join(clip, 'keypoints_new/person_1')+'/*.json'))
12
+
13
+ upper_body_points = list(np.arange(0, 25))
14
+ poses = []
15
+ confs = []
16
+ neck_to_nose_len = []
17
+ mean_position = []
18
+ for kp_file in keypoints_files:
19
+ kp_load = json.load(open(kp_file, 'r'))['people'][0]
20
+ posepts = kp_load['pose_keypoints_2d']
21
+ lhandpts = kp_load['hand_left_keypoints_2d']
22
+ rhandpts = kp_load['hand_right_keypoints_2d']
23
+ facepts = kp_load['face_keypoints_2d']
24
+
25
+ neck = np.array(posepts).reshape(-1,3)[1]
26
+ nose = np.array(posepts).reshape(-1,3)[0]
27
+ x_offset = abs(neck[0]-nose[0])
28
+ y_offset = abs(neck[1]-nose[1])
29
+ neck_to_nose_len.append(y_offset)
30
+ mean_position.append([neck[0],neck[1]])
31
+
32
+ keypoints=np.array(posepts+lhandpts+rhandpts+facepts).reshape(-1,3)[:,:2]
33
+
34
+ upper_body = keypoints[upper_body_points, :]
35
+ hand_points = keypoints[25:, :]
36
+ keypoints = np.vstack([upper_body, hand_points])
37
+
38
+ poses.append(keypoints)
39
+
40
+ if len(neck_to_nose_len) > 0:
41
+ scale_factor = np.mean(neck_to_nose_len)
42
+ else:
43
+ raise ValueError(clip)
44
+ mean_position = np.mean(np.array(mean_position), axis=0)
45
+
46
+ unlocalized_poses = np.array(poses).copy()
47
+ localized_poses = []
48
+ for i in range(len(poses)):
49
+ keypoints = poses[i]
50
+ neck = keypoints[1].copy()
51
+
52
+ keypoints[:, 0] = (keypoints[:, 0] - neck[0]) / scale_factor
53
+ keypoints[:, 1] = (keypoints[:, 1] - neck[1]) / scale_factor
54
+ localized_poses.append(keypoints.reshape(-1))
55
+
56
+ localized_poses=np.array(localized_poses)
57
+ return unlocalized_poses, localized_poses, (scale_factor, mean_position)
58
+
59
+ def get_full_path(wav_name, speaker, split):
60
+ '''
61
+ get clip path from aud file
62
+ '''
63
+ wav_name = os.path.basename(wav_name)
64
+ wav_name = os.path.splitext(wav_name)[0]
65
+ clip_name, vid_name = wav_name[:10], wav_name[11:]
66
+
67
+ full_path = os.path.join('pose_dataset/videos/', speaker, 'clips', vid_name, 'images/half', split, clip_name)
68
+
69
+ assert os.path.isdir(full_path), full_path
70
+
71
+ return full_path
72
+
73
+ def smooth(res):
74
+ '''
75
+ res: (B, seq_len, pose_dim)
76
+ '''
77
+ window = [res[:, 7, :], res[:, 8, :], res[:, 9, :], res[:, 10, :], res[:, 11, :], res[:, 12, :]]
78
+ w_size=7
79
+ for i in range(10, res.shape[1]-3):
80
+ window.append(res[:, i+3, :])
81
+ if len(window) > w_size:
82
+ window = window[1:]
83
+
84
+ if (i%25) in [22, 23, 24, 0, 1, 2, 3]:
85
+ res[:, i, :] = np.mean(window, axis=1)
86
+
87
+ return res
88
+
89
+ def cvt25(pred_poses, gt_poses=None):
90
+ '''
91
+ gt_poses: (1, seq_len, 270), 135 *2
92
+ pred_poses: (B, seq_len, 108), 54 * 2
93
+ '''
94
+ if gt_poses is None:
95
+ gt_poses = np.zeros_like(pred_poses)
96
+ else:
97
+ gt_poses = gt_poses.repeat(pred_poses.shape[0], axis=0)
98
+
99
+ length = min(pred_poses.shape[1], gt_poses.shape[1])
100
+ pred_poses = pred_poses[:, :length, :]
101
+ gt_poses = gt_poses[:, :length, :]
102
+ gt_poses = gt_poses.reshape(gt_poses.shape[0], gt_poses.shape[1], -1, 2)
103
+ pred_poses = pred_poses.reshape(pred_poses.shape[0], pred_poses.shape[1], -1, 2)
104
+
105
+ gt_poses[:, :, [1, 2, 3, 4, 5, 6, 7], :] = pred_poses[:, :, 1:8, :]
106
+ gt_poses[:, :, 25:25+21+21, :] = pred_poses[:, :, 12:, :]
107
+
108
+ return gt_poses.reshape(gt_poses.shape[0], gt_poses.shape[1], -1)
109
+
110
+ def hand_points(seq):
111
+ '''
112
+ seq: (B, seq_len, 135*2)
113
+ hands only
114
+ '''
115
+ hand_idx = [1, 2, 3, 4,5 ,6,7] + list(range(25, 25+21+21))
116
+ seq = seq.reshape(seq.shape[0], seq.shape[1], -1, 2)
117
+ return seq[:, :, hand_idx, :].reshape(seq.shape[0], seq.shape[1], -1)
118
+
119
+ def valid_points(seq):
120
+ '''
121
+ hands with some head points
122
+ '''
123
+ valid_idx = [0, 1, 2, 3, 4,5 ,6,7, 8, 9, 10, 11] + list(range(25, 25+21+21))
124
+ seq = seq.reshape(seq.shape[0], seq.shape[1], -1, 2)
125
+
126
+ seq = seq[:, :, valid_idx, :].reshape(seq.shape[0], seq.shape[1], -1)
127
+ assert seq.shape[-1] == 108, seq.shape
128
+ return seq
129
+
130
+ def draw_cdf(seq, save_name='cdf.jpg', color='slatebule'):
131
+ plt.figure()
132
+ plt.hist(seq, bins=100, range=(0, 100), color=color)
133
+ plt.savefig(save_name)
134
+
135
+ def to_excel(seq, save_name='res.xlsx'):
136
+ '''
137
+ seq: (T)
138
+ '''
139
+ df = pd.DataFrame(seq)
140
+ writer = pd.ExcelWriter(save_name)
141
+ df.to_excel(writer, 'sheet1')
142
+ writer.save()
143
+ writer.close()
144
+
145
+
146
+ if __name__ == '__main__':
147
+ random_data = np.random.randint(0, 10, 100)
148
+ draw_cdf(random_data)
losses/__init__.py ADDED
@@ -0,0 +1 @@
 
 
1
+ from .losses import *
losses/__pycache__/__init__.cpython-37.pyc ADDED
Binary file (174 Bytes). View file
 
losses/__pycache__/losses.cpython-37.pyc ADDED
Binary file (3.53 kB). View file
 
losses/losses.py ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import sys
3
+
4
+ sys.path.append(os.getcwd())
5
+
6
+ import torch
7
+ import torch.nn as nn
8
+ import torch.nn.functional as F
9
+ import numpy as np
10
+
11
+ class KeypointLoss(nn.Module):
12
+ def __init__(self):
13
+ super(KeypointLoss, self).__init__()
14
+
15
+ def forward(self, pred_seq, gt_seq, gt_conf=None):
16
+ #pred_seq: (B, C, T)
17
+ if gt_conf is not None:
18
+ gt_conf = gt_conf >= 0.01
19
+ return F.mse_loss(pred_seq[gt_conf], gt_seq[gt_conf], reduction='mean')
20
+ else:
21
+ return F.mse_loss(pred_seq, gt_seq)
22
+
23
+
24
+ class KLLoss(nn.Module):
25
+ def __init__(self, kl_tolerance):
26
+ super(KLLoss, self).__init__()
27
+ self.kl_tolerance = kl_tolerance
28
+
29
+ def forward(self, mu, var, mul=1):
30
+ kl_tolerance = self.kl_tolerance * mul * var.shape[1] / 64
31
+ kld_loss = -0.5 * torch.sum(1 + var - mu**2 - var.exp(), dim=1)
32
+ # kld_loss = -0.5 * torch.sum(1 + (var-1) - (mu) ** 2 - (var-1).exp(), dim=1)
33
+ if self.kl_tolerance is not None:
34
+ # above_line = kld_loss[kld_loss > self.kl_tolerance]
35
+ # if len(above_line) > 0:
36
+ # kld_loss = torch.mean(kld_loss)
37
+ # else:
38
+ # kld_loss = 0
39
+ kld_loss = torch.where(kld_loss > kl_tolerance, kld_loss, torch.tensor(kl_tolerance, device='cuda'))
40
+ # else:
41
+ kld_loss = torch.mean(kld_loss)
42
+ return kld_loss
43
+
44
+
45
+ class L2KLLoss(nn.Module):
46
+ def __init__(self, kl_tolerance):
47
+ super(L2KLLoss, self).__init__()
48
+ self.kl_tolerance = kl_tolerance
49
+
50
+ def forward(self, x):
51
+ # TODO: check
52
+ kld_loss = torch.sum(x ** 2, dim=1)
53
+ if self.kl_tolerance is not None:
54
+ above_line = kld_loss[kld_loss > self.kl_tolerance]
55
+ if len(above_line) > 0:
56
+ kld_loss = torch.mean(kld_loss)
57
+ else:
58
+ kld_loss = 0
59
+ else:
60
+ kld_loss = torch.mean(kld_loss)
61
+ return kld_loss
62
+
63
+ class L2RegLoss(nn.Module):
64
+ def __init__(self):
65
+ super(L2RegLoss, self).__init__()
66
+
67
+ def forward(self, x):
68
+ #TODO: check
69
+ return torch.sum(x**2)
70
+
71
+
72
+ class L2Loss(nn.Module):
73
+ def __init__(self):
74
+ super(L2Loss, self).__init__()
75
+
76
+ def forward(self, x):
77
+ # TODO: check
78
+ return torch.sum(x ** 2)
79
+
80
+
81
+ class AudioLoss(nn.Module):
82
+ def __init__(self):
83
+ super(AudioLoss, self).__init__()
84
+
85
+ def forward(self, dynamics, gt_poses):
86
+ #pay attention, normalized
87
+ mean = torch.mean(gt_poses, dim=-1).unsqueeze(-1)
88
+ gt = gt_poses - mean
89
+ return F.mse_loss(dynamics, gt)
90
+
91
+ L1Loss = nn.L1Loss
nets/LS3DCG.py ADDED
@@ -0,0 +1,414 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ '''
2
+ not exactly the same as the official repo but the results are good
3
+ '''
4
+ import sys
5
+ import os
6
+
7
+ from data_utils.lower_body import c_index_3d, c_index_6d
8
+
9
+ sys.path.append(os.getcwd())
10
+
11
+ import numpy as np
12
+ import torch
13
+ import torch.nn as nn
14
+ import torch.optim as optim
15
+ import torch.nn.functional as F
16
+ import math
17
+
18
+ from nets.base import TrainWrapperBaseClass
19
+ from nets.layers import SeqEncoder1D
20
+ from losses import KeypointLoss, L1Loss, KLLoss
21
+ from data_utils.utils import get_melspec, get_mfcc_psf, get_mfcc_ta
22
+ from nets.utils import denormalize
23
+
24
+ class Conv1d_tf(nn.Conv1d):
25
+ """
26
+ Conv1d with the padding behavior from TF
27
+ modified from https://github.com/mlperf/inference/blob/482f6a3beb7af2fb0bd2d91d6185d5e71c22c55f/others/edge/object_detection/ssd_mobilenet/pytorch/utils.py
28
+ """
29
+
30
+ def __init__(self, *args, **kwargs):
31
+ super(Conv1d_tf, self).__init__(*args, **kwargs)
32
+ self.padding = kwargs.get("padding", "same")
33
+
34
+ def _compute_padding(self, input, dim):
35
+ input_size = input.size(dim + 2)
36
+ filter_size = self.weight.size(dim + 2)
37
+ effective_filter_size = (filter_size - 1) * self.dilation[dim] + 1
38
+ out_size = (input_size + self.stride[dim] - 1) // self.stride[dim]
39
+ total_padding = max(
40
+ 0, (out_size - 1) * self.stride[dim] + effective_filter_size - input_size
41
+ )
42
+ additional_padding = int(total_padding % 2 != 0)
43
+
44
+ return additional_padding, total_padding
45
+
46
+ def forward(self, input):
47
+ if self.padding == "VALID":
48
+ return F.conv1d(
49
+ input,
50
+ self.weight,
51
+ self.bias,
52
+ self.stride,
53
+ padding=0,
54
+ dilation=self.dilation,
55
+ groups=self.groups,
56
+ )
57
+ rows_odd, padding_rows = self._compute_padding(input, dim=0)
58
+ if rows_odd:
59
+ input = F.pad(input, [0, rows_odd])
60
+
61
+ return F.conv1d(
62
+ input,
63
+ self.weight,
64
+ self.bias,
65
+ self.stride,
66
+ padding=(padding_rows // 2),
67
+ dilation=self.dilation,
68
+ groups=self.groups,
69
+ )
70
+
71
+
72
+ def ConvNormRelu(in_channels, out_channels, type='1d', downsample=False, k=None, s=None, norm='bn', padding='valid'):
73
+ if k is None and s is None:
74
+ if not downsample:
75
+ k = 3
76
+ s = 1
77
+ else:
78
+ k = 4
79
+ s = 2
80
+
81
+ if type == '1d':
82
+ conv_block = Conv1d_tf(in_channels, out_channels, kernel_size=k, stride=s, padding=padding)
83
+ if norm == 'bn':
84
+ norm_block = nn.BatchNorm1d(out_channels)
85
+ elif norm == 'ln':
86
+ norm_block = nn.LayerNorm(out_channels)
87
+ elif type == '2d':
88
+ conv_block = Conv2d_tf(in_channels, out_channels, kernel_size=k, stride=s, padding=padding)
89
+ norm_block = nn.BatchNorm2d(out_channels)
90
+ else:
91
+ assert False
92
+
93
+ return nn.Sequential(
94
+ conv_block,
95
+ norm_block,
96
+ nn.LeakyReLU(0.2, True)
97
+ )
98
+
99
+ class Decoder(nn.Module):
100
+ def __init__(self, in_ch, out_ch):
101
+ super(Decoder, self).__init__()
102
+ self.up1 = nn.Sequential(
103
+ ConvNormRelu(in_ch // 2 + in_ch, in_ch // 2),
104
+ ConvNormRelu(in_ch // 2, in_ch // 2),
105
+ nn.Upsample(scale_factor=2, mode='nearest')
106
+ )
107
+ self.up2 = nn.Sequential(
108
+ ConvNormRelu(in_ch // 4 + in_ch // 2, in_ch // 4),
109
+ ConvNormRelu(in_ch // 4, in_ch // 4),
110
+ nn.Upsample(scale_factor=2, mode='nearest')
111
+ )
112
+ self.up3 = nn.Sequential(
113
+ ConvNormRelu(in_ch // 8 + in_ch // 4, in_ch // 8),
114
+ ConvNormRelu(in_ch // 8, in_ch // 8),
115
+ nn.Conv1d(in_ch // 8, out_ch, 1, 1)
116
+ )
117
+
118
+ def forward(self, x, x1, x2, x3):
119
+ x = F.interpolate(x, x3.shape[2])
120
+ x = torch.cat([x, x3], dim=1)
121
+ x = self.up1(x)
122
+ x = F.interpolate(x, x2.shape[2])
123
+ x = torch.cat([x, x2], dim=1)
124
+ x = self.up2(x)
125
+ x = F.interpolate(x, x1.shape[2])
126
+ x = torch.cat([x, x1], dim=1)
127
+ x = self.up3(x)
128
+ return x
129
+
130
+
131
+ class EncoderDecoder(nn.Module):
132
+ def __init__(self, n_frames, each_dim):
133
+ super().__init__()
134
+ self.n_frames = n_frames
135
+
136
+ self.down1 = nn.Sequential(
137
+ ConvNormRelu(64, 64, '1d', False),
138
+ ConvNormRelu(64, 128, '1d', False),
139
+ )
140
+ self.down2 = nn.Sequential(
141
+ ConvNormRelu(128, 128, '1d', False),
142
+ ConvNormRelu(128, 256, '1d', False),
143
+ )
144
+ self.down3 = nn.Sequential(
145
+ ConvNormRelu(256, 256, '1d', False),
146
+ ConvNormRelu(256, 512, '1d', False),
147
+ )
148
+ self.down4 = nn.Sequential(
149
+ ConvNormRelu(512, 512, '1d', False),
150
+ ConvNormRelu(512, 1024, '1d', False),
151
+ )
152
+
153
+ self.down = nn.MaxPool1d(kernel_size=2)
154
+ self.up = nn.Upsample(scale_factor=2, mode='nearest')
155
+
156
+ self.face_decoder = Decoder(1024, each_dim[0] + each_dim[3])
157
+ self.body_decoder = Decoder(1024, each_dim[1])
158
+ self.hand_decoder = Decoder(1024, each_dim[2])
159
+
160
+ def forward(self, spectrogram, time_steps=None):
161
+ if time_steps is None:
162
+ time_steps = self.n_frames
163
+
164
+ x1 = self.down1(spectrogram)
165
+ x = self.down(x1)
166
+ x2 = self.down2(x)
167
+ x = self.down(x2)
168
+ x3 = self.down3(x)
169
+ x = self.down(x3)
170
+ x = self.down4(x)
171
+ x = self.up(x)
172
+
173
+ face = self.face_decoder(x, x1, x2, x3)
174
+ body = self.body_decoder(x, x1, x2, x3)
175
+ hand = self.hand_decoder(x, x1, x2, x3)
176
+
177
+ return face, body, hand
178
+
179
+
180
+ class Generator(nn.Module):
181
+ def __init__(self,
182
+ each_dim,
183
+ training=False,
184
+ device=None
185
+ ):
186
+ super().__init__()
187
+
188
+ self.training = training
189
+ self.device = device
190
+
191
+ self.encoderdecoder = EncoderDecoder(15, each_dim)
192
+
193
+ def forward(self, in_spec, time_steps=None):
194
+ if time_steps is not None:
195
+ self.gen_length = time_steps
196
+
197
+ face, body, hand = self.encoderdecoder(in_spec)
198
+ out = torch.cat([face, body, hand], dim=1)
199
+ out = out.transpose(1, 2)
200
+
201
+ return out
202
+
203
+
204
+ class Discriminator(nn.Module):
205
+ def __init__(self, input_dim):
206
+ super().__init__()
207
+ self.net = nn.Sequential(
208
+ ConvNormRelu(input_dim, 128, '1d'),
209
+ ConvNormRelu(128, 256, '1d'),
210
+ nn.MaxPool1d(kernel_size=2),
211
+ ConvNormRelu(256, 256, '1d'),
212
+ ConvNormRelu(256, 512, '1d'),
213
+ nn.MaxPool1d(kernel_size=2),
214
+ ConvNormRelu(512, 512, '1d'),
215
+ ConvNormRelu(512, 1024, '1d'),
216
+ nn.MaxPool1d(kernel_size=2),
217
+ nn.Conv1d(1024, 1, 1, 1),
218
+ nn.Sigmoid()
219
+ )
220
+
221
+ def forward(self, x):
222
+ x = x.transpose(1, 2)
223
+
224
+ out = self.net(x)
225
+ return out
226
+
227
+
228
+ class TrainWrapper(TrainWrapperBaseClass):
229
+ def __init__(self, args, config) -> None:
230
+ self.args = args
231
+ self.config = config
232
+ self.device = torch.device(self.args.gpu)
233
+ self.global_step = 0
234
+ self.convert_to_6d = self.config.Data.pose.convert_to_6d
235
+ self.init_params()
236
+
237
+ self.generator = Generator(
238
+ each_dim=self.each_dim,
239
+ training=not self.args.infer,
240
+ device=self.device,
241
+ ).to(self.device)
242
+ self.discriminator = Discriminator(
243
+ input_dim=self.each_dim[1] + self.each_dim[2] + 64
244
+ ).to(self.device)
245
+ if self.convert_to_6d:
246
+ self.c_index = c_index_6d
247
+ else:
248
+ self.c_index = c_index_3d
249
+ self.MSELoss = KeypointLoss().to(self.device)
250
+ self.L1Loss = L1Loss().to(self.device)
251
+ super().__init__(args, config)
252
+
253
+ def init_params(self):
254
+ scale = 1
255
+
256
+ global_orient = round(0 * scale)
257
+ leye_pose = reye_pose = round(0 * scale)
258
+ jaw_pose = round(3 * scale)
259
+ body_pose = round((63 - 24) * scale)
260
+ left_hand_pose = right_hand_pose = round(45 * scale)
261
+
262
+ expression = 100
263
+
264
+ b_j = 0
265
+ jaw_dim = jaw_pose
266
+ b_e = b_j + jaw_dim
267
+ eye_dim = leye_pose + reye_pose
268
+ b_b = b_e + eye_dim
269
+ body_dim = global_orient + body_pose
270
+ b_h = b_b + body_dim
271
+ hand_dim = left_hand_pose + right_hand_pose
272
+ b_f = b_h + hand_dim
273
+ face_dim = expression
274
+
275
+ self.dim_list = [b_j, b_e, b_b, b_h, b_f]
276
+ self.full_dim = jaw_dim + eye_dim + body_dim + hand_dim
277
+ self.pose = int(self.full_dim / round(3 * scale))
278
+ self.each_dim = [jaw_dim, eye_dim + body_dim, hand_dim, face_dim]
279
+
280
+ def __call__(self, bat):
281
+ assert (not self.args.infer), "infer mode"
282
+ self.global_step += 1
283
+
284
+ loss_dict = {}
285
+
286
+ aud, poses = bat['aud_feat'].to(self.device).to(torch.float32), bat['poses'].to(self.device).to(torch.float32)
287
+ expression = bat['expression'].to(self.device).to(torch.float32)
288
+ jaw = poses[:, :3, :]
289
+ poses = poses[:, self.c_index, :]
290
+
291
+ pred = self.generator(in_spec=aud)
292
+
293
+ D_loss, D_loss_dict = self.get_loss(
294
+ pred_poses=pred.detach(),
295
+ gt_poses=poses,
296
+ aud=aud,
297
+ mode='training_D',
298
+ )
299
+
300
+ self.discriminator_optimizer.zero_grad()
301
+ D_loss.backward()
302
+ self.discriminator_optimizer.step()
303
+
304
+ G_loss, G_loss_dict = self.get_loss(
305
+ pred_poses=pred,
306
+ gt_poses=poses,
307
+ aud=aud,
308
+ expression=expression,
309
+ jaw=jaw,
310
+ mode='training_G',
311
+ )
312
+ self.generator_optimizer.zero_grad()
313
+ G_loss.backward()
314
+ self.generator_optimizer.step()
315
+
316
+ total_loss = None
317
+ loss_dict = {}
318
+ for key in list(D_loss_dict.keys()) + list(G_loss_dict.keys()):
319
+ loss_dict[key] = G_loss_dict.get(key, 0) + D_loss_dict.get(key, 0)
320
+
321
+ return total_loss, loss_dict
322
+
323
+ def get_loss(self,
324
+ pred_poses,
325
+ gt_poses,
326
+ aud=None,
327
+ jaw=None,
328
+ expression=None,
329
+ mode='training_G',
330
+ ):
331
+ loss_dict = {}
332
+ aud = aud.transpose(1, 2)
333
+ gt_poses = gt_poses.transpose(1, 2)
334
+ gt_aud = torch.cat([gt_poses, aud], dim=2)
335
+ pred_aud = torch.cat([pred_poses[:, :, 103:], aud], dim=2)
336
+
337
+ if mode == 'training_D':
338
+ dis_real = self.discriminator(gt_aud)
339
+ dis_fake = self.discriminator(pred_aud)
340
+ dis_error = self.MSELoss(torch.ones_like(dis_real).to(self.device), dis_real) + self.MSELoss(
341
+ torch.zeros_like(dis_fake).to(self.device), dis_fake)
342
+ loss_dict['dis'] = dis_error
343
+
344
+ return dis_error, loss_dict
345
+ elif mode == 'training_G':
346
+ jaw_loss = self.L1Loss(pred_poses[:, :, :3], jaw.transpose(1, 2))
347
+ face_loss = self.MSELoss(pred_poses[:, :, 3:103], expression.transpose(1, 2))
348
+ body_loss = self.L1Loss(pred_poses[:, :, 103:142], gt_poses[:, :, :39])
349
+ hand_loss = self.L1Loss(pred_poses[:, :, 142:], gt_poses[:, :, 39:])
350
+ l1_loss = jaw_loss + face_loss + body_loss + hand_loss
351
+
352
+ dis_output = self.discriminator(pred_aud)
353
+ gen_error = self.MSELoss(torch.ones_like(dis_output).to(self.device), dis_output)
354
+ gen_loss = self.config.Train.weights.keypoint_loss_weight * l1_loss + self.config.Train.weights.gan_loss_weight * gen_error
355
+
356
+ loss_dict['gen'] = gen_error
357
+ loss_dict['jaw_loss'] = jaw_loss
358
+ loss_dict['face_loss'] = face_loss
359
+ loss_dict['body_loss'] = body_loss
360
+ loss_dict['hand_loss'] = hand_loss
361
+ return gen_loss, loss_dict
362
+ else:
363
+ raise ValueError(mode)
364
+
365
+ def infer_on_audio(self, aud_fn, fps=30, initial_pose=None, norm_stats=None, id=None, B=1, **kwargs):
366
+ output = []
367
+ assert self.args.infer, "train mode"
368
+ self.generator.eval()
369
+
370
+ if self.config.Data.pose.normalization:
371
+ assert norm_stats is not None
372
+ data_mean = norm_stats[0]
373
+ data_std = norm_stats[1]
374
+
375
+ pre_length = self.config.Data.pose.pre_pose_length
376
+ generate_length = self.config.Data.pose.generate_length
377
+ # assert pre_length == initial_pose.shape[-1]
378
+ # pre_poses = initial_pose.permute(0, 2, 1).to(self.device).to(torch.float32)
379
+ # B = pre_poses.shape[0]
380
+
381
+ aud_feat = get_mfcc_ta(aud_fn, sr=22000, fps=fps, smlpx=True, type='mfcc').transpose(1, 0)
382
+ num_poses_to_generate = aud_feat.shape[-1]
383
+ aud_feat = aud_feat[np.newaxis, ...].repeat(B, axis=0)
384
+ aud_feat = torch.tensor(aud_feat, dtype=torch.float32).to(self.device)
385
+
386
+ with torch.no_grad():
387
+ pred_poses = self.generator(aud_feat)
388
+ pred_poses = pred_poses.cpu().numpy()
389
+ output = pred_poses.squeeze()
390
+
391
+ return output
392
+
393
+ def generate(self, aud, id):
394
+ self.generator.eval()
395
+ pred_poses = self.generator(aud)
396
+ return pred_poses
397
+
398
+
399
+ if __name__ == '__main__':
400
+ from trainer.options import parse_args
401
+
402
+ parser = parse_args()
403
+ args = parser.parse_args(
404
+ ['--exp_name', '0', '--data_root', '0', '--speakers', '0', '--pre_pose_length', '4', '--generate_length', '64',
405
+ '--infer'])
406
+
407
+ generator = TrainWrapper(args)
408
+
409
+ aud_fn = '../sample_audio/jon.wav'
410
+ initial_pose = torch.randn(64, 108, 4)
411
+ norm_stats = (np.random.randn(108), np.random.randn(108))
412
+ output = generator.infer_on_audio(aud_fn, initial_pose, norm_stats)
413
+
414
+ print(output.shape)
nets/__init__.py ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ from .smplx_face import TrainWrapper as s2g_face
2
+ from .smplx_body_vq import TrainWrapper as s2g_body_vq
3
+ from .smplx_body_pixel import TrainWrapper as s2g_body_pixel
4
+ from .body_ae import TrainWrapper as s2g_body_ae
5
+ from .LS3DCG import TrainWrapper as LS3DCG
6
+ from .base import TrainWrapperBaseClass
7
+
8
+ from .utils import normalize, denormalize
nets/__pycache__/__init__.cpython-37.pyc ADDED
Binary file (407 Bytes). View file
 
nets/__pycache__/base.cpython-37.pyc ADDED
Binary file (2.29 kB). View file
 
nets/__pycache__/init_model.cpython-37.pyc ADDED
Binary file (460 Bytes). View file