Update README.md

import cv2
import os
import numpy as np
import random
import joblib
from sklearn import svm
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA


class FaceRecognizer:
def __init__(self, imgdir='./test_img/', grayfacedir='./test_img_gray', model_path='./svm_model.pkl'):
self.imgdir = imgdir
self.grayfacedir = grayfacedir
self.model_path = model_path
self.svm_model = svm.SVC(kernel='linear', probability=True)
self.face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
self.label_map = {}
self.pca_model = None # 初始化 pca_model 为 None


def collect_data(self, someone, picturenum=50):
"""
收集人脸数据并保存图像，同时进行数据增强
"""
person_dir = self._create_person_directory(someone)
capture = cv2.VideoCapture(0)

cv2.waitKey(1)

count = 0
while count < picturenum:
ret, img = capture.read()
if not ret:
print("无法获取图像")
break

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
faces = self.detect_faces(gray)

for (x, y, w, h) in faces:
face_img = gray[y:y + h, x:x + w]
augmented_faces = self.augment_face(face_img)

for augmented_face in augmented_faces:
picturepath = os.path.join(person_dir, f'{count + 1}.jpg')
cv2.imwrite(picturepath, augmented_face)
count += 1

if count >= picturenum:
break

cv2.imshow(f'Capturing Face - {someone}', img)

if cv2.waitKey(100) == ord('q'):
break

capture.release()
cv2.destroyAllWindows()

def _create_person_directory(self, someone):
"""
创建人物名称的子文件夹
"""
person_dir = os.path.join(self.imgdir, someone)
if not os.path.exists(person_dir):
os.makedirs(person_dir)
return person_dir

def detect_faces(self, gray):
"""
检测图像中的人脸
"""
faces = self.face_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
return faces

def augment_face(self, face_img):
"""
对人脸图像进行数据增强，包括旋转、镜像、平移等
"""
augmented_faces = [cv2.flip(face_img, 1)] # 镜像（水平翻转）

angle = random.uniform(-10, 10) # 随机旋转
augmented_faces.append(self.rotate_image(face_img, angle))

translation = random.randint(-5, 5) # 随机平移
augmented_faces.append(self.translate_image(face_img, translation))

return augmented_faces

def rotate_image(self, image, angle):
"""
旋转图像指定角度
"""
height, width = image.shape
center = (width // 2, height // 2)
rotation_matrix = cv2.getRotationMatrix2D(center, angle, 1.0)
return cv2.warpAffine(image, rotation_matrix, (width, height))

def translate_image(self, image, tx):
"""
图像平移
"""
height, width = image.shape
translation_matrix = np.float32([[1, 0, tx], [0, 1, 0]])
return cv2.warpAffine(image, translation_matrix, (width, height))

def prepare_data(self):
"""
准备训练数据并进行标准化处理
"""
faces = []
labels = []

person_id = 0
for person_name in os.listdir(self.imgdir):
person_dir = os.path.join(self.imgdir, person_name)
if not os.path.isdir(person_dir):
continue

if person_name not in self.label_map:
self.label_map[person_id] = person_name # 新人物，分配新的ID
person_id += 1

for img_name in os.listdir(person_dir):
img_path = os.path.join(person_dir, img_name)
if img_name.lower().endswith(('.png', '.jpg', '.jpeg')):
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

faces_detected = self.detect_faces(gray)
for (x, y, w, h) in faces_detected:
face = gray[y:y + h, x:x + w]
face_resized = cv2.resize(face, (100, 100))
faces.append(face_resized)
labels.append(self.get_person_id(person_name))

faces = np.array(faces)
faces = faces.reshape(faces.shape[0], -1) # 展平图像数据
faces = StandardScaler().fit_transform(faces) # 进行标准化处理

return faces, labels

def train_svm(self, n_components=50):
"""
使用 PCA 进行降维后，使用 SVM 训练模型
"""
faces, labels = self.prepare_data()
self.pca = PCA(n_components=n_components, whiten=True)
faces_pca = self.pca.fit_transform(faces)

self.svm_model.fit(faces_pca, labels)

self.save_model()
joblib.dump(self.pca, './pca_model.pkl')
print("SVM 模型和 PCA 降维模型训练完成！")

def get_person_id(self, person_name):
"""
根据人物名称获取对应的ID，如果没有，则让用户命名
"""
if person_name not in self.label_map.values():
new_name = input(f"请输入 {person_name} 的名称：")
person_id = len(self.label_map)
self.label_map[person_id] = new_name
print(f"已将 {person_name} 命名为 {new_name}")
return person_id
return list(self.label_map.values()).index(person_name)

def save_model(self):
"""
保存训练好的SVM模型
"""
joblib.dump(self.svm_model, self.model_path)
print(f"模型已保存至 {self.model_path}")

def load_model(self):
"""
加载已保存的SVM模型和PCA模型
"""
self.svm_model = joblib.load(self.model_path)
self.pca_model = joblib.load('./pca_model.pkl') # 加载PCA模型
print(f"模型已加载：{self.model_path} 和 ./pca_model.pkl")

def predict(self, img):
"""
使用训练好的SVM模型进行人脸识别，并应用PCA降维
"""
if self.pca_model is None: # 检查 pca_model 是否已加载
print("PCA模型尚未加载，请先加载模型。")
return

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
faces_detected = self.face_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
for (x, y, w, h) in faces_detected:
face = gray[y:y + h, x:x + w]
face_resized = cv2.resize(face, (100, 100)) # 调整大小
face_flattened = face_resized.flatten().reshape(1, -1)
face_flattened = StandardScaler().fit_transform(face_flattened)
face_pca = self.pca_model.transform(face_flattened) # 使用训练时的PCA进行降维
label = self.svm_model.predict(face_pca)
confidence = self.svm_model.decision_function(face_pca)
person_name = self.label_map.get(label[0], 'Unknown')
print(f"预测标签：{person_name}, 置信度：{confidence}")
cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)
cv2.putText(img, person_name, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.9, (0, 255, 0), 2)


def prepare_data(self):
"""
准备训练数据并进行标准化处理
"""
faces = []
labels = []

person_id = 0
for person_name in os.listdir(self.imgdir):
person_dir = os.path.join(self.imgdir, person_name)
if not os.path.isdir(person_dir):
continue

if person_name not in self.label_map:
self.label_map[person_id] = person_name # 新人物，分配新的ID
person_id += 1

# 处理每个人物的图像
for img_name in os.listdir(person_dir):
img_path = os.path.join(person_dir, img_name)
if img_name.lower().endswith(('.png', '.jpg', '.jpeg')):
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# 进行人脸检测
faces_detected = self.face_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5,
minSize=(30, 30))
for (x, y, w, h) in faces_detected:
face = gray[y:y + h, x:x + w]
face_resized = cv2.resize(face, (100, 100)) # 统一调整大小
faces.append(face_resized)
labels.append(self.get_person_id(person_name))

# 标准化
faces = np.array(faces)
faces = faces.reshape(faces.shape[0], -1) # 展平图像数据
faces = StandardScaler().fit_transform(faces) # 进行标准化处理

return faces, labels
def train_svm(self, n_components=50):
"""
使用 PCA 进行降维后，使用 SVM 训练模型
"""
faces, labels = self.prepare_data()

# 使用 PCA 进行降维
self.pca = PCA(n_components=n_components, whiten=True)
faces_pca = self.pca.fit_transform(faces)

# 使用 SVM 进行训练
self.svm_model.fit(faces_pca, labels)

# 保存模型和 PCA 对象
self.save_model()
joblib.dump(self.pca, './pca_model.pkl')
print("SVM 模型和 PCA 降维模型训练完成！")

def get_person_id(self, person_name):
"""
根据人物名称获取对应的ID，如果没有，则让用户命名
"""
# 如果人物名称不存在，提示用户输入新的名称

README.md

@@ -1,3 +1,10 @@
+---
+license: apache-2.0
+datasets:
+- HuggingFaceFW/fineweb-2
+language:
+- aa
+---
 <!-- markdownlint-disable first-line-h1 -->
 <!-- markdownlint-disable html -->
 <!-- markdownlint-disable no-duplicate-header -->
@@ -333,4 +340,4 @@ This code repository is licensed under [the MIT License](LICENSE-CODE). The use
 ```
 
 ## 9. Contact
-If you have any questions, please raise an issue or contact us at [[email protected]]([email protected]).
+If you have any questions, please raise an issue or contact us at [[email protected]]([email protected]).