Stavros Niafas commited on
Commit
67c3c5b
·
1 Parent(s): 142a9d0

update model card

Browse files
Files changed (1) hide show
  1. README.md +28 -2
README.md CHANGED
@@ -2,10 +2,36 @@
2
  license: apache-2.0
3
  tags:
4
  - vision
5
- - classification
 
6
  datasets:
7
- - svnfs/depth_of_field
 
 
 
 
 
8
  ---
9
  # Bokeh (ボケ Japanese word for blur)
10
 
11
  Bokeh model is based on a densenet like architecture trained on Unsplash images at 300x200 resolution. It classifies whether an photo is capture with bokeh producing a shallow depth of field
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
  license: apache-2.0
3
  tags:
4
  - vision
5
+ - image-classification
6
+ library_name: Keras
7
  datasets:
8
+ - svnfs/depth-of-field
9
+ widget:
10
+ - src: https://huggingface.co/datasets/svnfs/depth-of-field/blob/main/data/0/-1a83VD65ss.jpg
11
+ example_title: Shallow DoF
12
+ - src: https://huggingface.co/datasets/svnfs/depth-of-field/blob/main/data/1/007R8JewpwU.jpg
13
+ example_title: Deep DoF
14
  ---
15
  # Bokeh (ボケ Japanese word for blur)
16
 
17
  Bokeh model is based on a densenet like architecture trained on Unsplash images at 300x200 resolution. It classifies whether an photo is capture with bokeh producing a shallow depth of field
18
+
19
+ ## Model description
20
+
21
+ Bokeh model is based on a DenseNet architecture. The model is trained with a mini-batch size of 32 samples with Adam optimizer and a learning rate $0.0001$.
22
+ It has 3.632 trainable parameters, 8 convolution filters are used for the network's input, with $7\times7$ kernel size.
23
+
24
+ ## Training data
25
+
26
+ The bokeh model is pretrained on [depth-of-field](https://huggingface.co/datasets/svnfs/depth-of-field) dataset, a dataset consisted of 1200 images and 2 classes manually annotated.
27
+ ### BibTeX entry and citation info
28
+
29
+ ```
30
+ @article{sniafas2021,
31
+ title={DoF: An image dataset for depth of field classification},
32
+ author={Niafas, Stavros},
33
+ doi= {10.13140/RG.2.2.17217.89443},
34
+ url= {https://www.researchgate.net/publication/355917312_Photography_Style_Analysis_using_Machine_Learning}
35
+ year={2021}
36
+ }
37
+ ```