已开启
new: 新建文件 1班20.py #102
mmahao创建于 2025年12月29日
new: 新建文件 1班20.py #102
已开启
共 1 个文件变更+510-0
| @@ -0,0 +1,510 @@ | |||
| 1 | +# face_recognition_system.py | ||
| 2 | +import cv2 | ||
| 3 | +import numpy as np | ||
| 4 | +import os | ||
| 5 | +import pickle | ||
| 6 | +import time | ||
| 7 | +import dlib | ||
| 8 | +from sklearn.svm import SVC | ||
| 9 | +from sklearn.preprocessing import LabelEncoder | ||
| 10 | +from sklearn.model_selection import train_test_split | ||
| 11 | +from sklearn.metrics import classification_report | ||
| 12 | +import face_recognition | ||
| 13 | +from imutils import paths | ||
| 14 | +import warnings | ||
| 15 | +warnings.filterwarnings('ignore') | ||
| 16 | + | ||
| 17 | +class FaceRecognizer: | ||
| 18 | + def __init__(self, method='cnn', tolerance=0.6, model_save_path='models'): | ||
| 19 | + self.method = method | ||
| 20 | + self.tolerance = tolerance | ||
| 21 | + self.model_save_path = model_save_path | ||
| 22 | + | ||
| 23 | + # 创建必要的目录 | ||
| 24 | + os.makedirs(model_save_path, exist_ok=True) | ||
| 25 | + os.makedirs('dataset', exist_ok=True) | ||
| 26 | + os.makedirs('output', exist_ok=True) | ||
| 27 | + | ||
| 28 | + # 初始化变量 | ||
| 29 | + self.known_face_encodings = [] | ||
| 30 | + self.known_face_names = [] | ||
| 31 | + self.face_classifier = None | ||
| 32 | + self.label_encoder = LabelEncoder() | ||
| 33 | + | ||
| 34 | + def create_dataset(self, name, num_samples=20): | ||
| 35 | + print(f"[INFO] 开始采集 {name} 的人脸数据...") | ||
| 36 | + | ||
| 37 | + # 创建个人文件夹 | ||
| 38 | + person_path = os.path.join('dataset', name) | ||
| 39 | + os.makedirs(person_path, exist_ok=True) | ||
| 40 | + | ||
| 41 | + # 初始化摄像头 | ||
| 42 | + cap = cv2.VideoCapture(0) | ||
| 43 | + if not cap.isOpened(): | ||
| 44 | + print("[ERROR] 无法打开摄像头") | ||
| 45 | + return | ||
| 46 | + | ||
| 47 | + face_cascade = cv2.CascadeClassifier( | ||
| 48 | + cv2.data.haarcascades + 'haarcascade_frontalface_default.xml' | ||
| 49 | + ) | ||
| 50 | + | ||
| 51 | + count = 0 | ||
| 52 | + while count < num_samples: | ||
| 53 | + ret, frame = cap.read() | ||
| 54 | + if not ret: | ||
| 55 | + continue | ||
| 56 | + | ||
| 57 | + # 转换为灰度图 | ||
| 58 | + gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) | ||
| 59 | + | ||
| 60 | + # 人脸检测 | ||
| 61 | + faces = face_cascade.detectMultiScale( | ||
| 62 | + gray, | ||
| 63 | + scaleFactor=1.1, | ||
| 64 | + minNeighbors=5, | ||
| 65 | + minSize=(30, 30) | ||
| 66 | + ) | ||
| 67 | + | ||
| 68 | + for (x, y, w, h) in faces: | ||
| 69 | + # 绘制矩形框 | ||
| 70 | + cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2) | ||
| 71 | + | ||
| 72 | + # 保存人脸图片 | ||
| 73 | + face_img = frame[y:y+h, x:x+w] | ||
| 74 | + face_img = cv2.resize(face_img, (160, 160)) | ||
| 75 | + | ||
| 76 | + # 每隔5帧保存一次 | ||
| 77 | + if count % 5 == 0: | ||
| 78 | + img_path = os.path.join(person_path, f"{name}_{count}.jpg") | ||
| 79 | + cv2.imwrite(img_path, face_img) | ||
| 80 | + count += 1 | ||
| 81 | + print(f"[INFO] 已保存 {count}/{num_samples} 张图片") | ||
| 82 | + | ||
| 83 | + # 显示视频 | ||
| 84 | + cv2.putText(frame, f"采集进度: {count}/{num_samples}", | ||
| 85 | + (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 2) | ||
| 86 | + cv2.imshow('创建数据集 - 按ESC退出', frame) | ||
| 87 | + | ||
| 88 | + if cv2.waitKey(1) & 0xFF == 27: # ESC键退出 | ||
| 89 | + break | ||
| 90 | + | ||
| 91 | + cap.release() | ||
| 92 | + cv2.destroyAllWindows() | ||
| 93 | + print(f"[INFO] {name} 的数据采集完成") | ||
| 94 | + | ||
| 95 | + def extract_face_encodings(self, dataset_path='dataset'): | ||
| 96 | + image_paths = list(paths.list_images(dataset_path)) | ||
| 97 | + | ||
| 98 | + if len(image_paths) == 0: | ||
| 99 | + print("[ERROR] 数据集中没有找到图片") | ||
| 100 | + return [], [] | ||
| 101 | + | ||
| 102 | + encodings = [] | ||
| 103 | + names = [] | ||
| 104 | + | ||
| 105 | + for (i, image_path) in enumerate(image_paths): | ||
| 106 | + print(f"[INFO] 处理图片 {i+1}/{len(image_paths)}") | ||
| 107 | + | ||
| 108 | + # 从路径中提取人名 | ||
| 109 | + name = image_path.split(os.path.sep)[-2] | ||
| 110 | + | ||
| 111 | + # 加载图片 | ||
| 112 | + image = cv2.imread(image_path) | ||
| 113 | + rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) | ||
| 114 | + | ||
| 115 | + # 检测人脸 | ||
| 116 | + face_locations = face_recognition.face_locations( | ||
| 117 | + rgb, model=self.method | ||
| 118 | + ) | ||
| 119 | + | ||
| 120 | + if len(face_locations) > 0: | ||
| 121 | + # 提取人脸编码 | ||
| 122 | + face_encoding = face_recognition.face_encodings( | ||
| 123 | + rgb, face_locations | ||
| 124 | + )[0] | ||
| 125 | + | ||
| 126 | + encodings.append(face_encoding) | ||
| 127 | + names.append(name) | ||
| 128 | + | ||
| 129 | + print(f"[INFO] 成功提取 {len(encodings)} 个人脸编码") | ||
| 130 | + return encodings, names | ||
| 131 | + | ||
| 132 | + def train_classifier(self, save_model=True): | ||
| 133 | + # 提取人脸编码 | ||
| 134 | + encodings, names = self.extract_face_encodings() | ||
| 135 | + | ||
| 136 | + if len(encodings) == 0: | ||
| 137 | + print("[ERROR] 没有可用的数据进行训练") | ||
| 138 | + return | ||
| 139 | + | ||
| 140 | + # 编码标签 | ||
| 141 | + labels = self.label_encoder.fit_transform(names) | ||
| 142 | + | ||
| 143 | + # 使用SVM分类器 | ||
| 144 | + self.face_classifier = SVC(kernel='linear', probability=True, C=1.0) | ||
| 145 | + self.face_classifier.fit(encodings, labels) | ||
| 146 | + | ||
| 147 | + # 保存已知的人脸编码 | ||
| 148 | + self.known_face_encodings = encodings | ||
| 149 | + self.known_face_names = names | ||
| 150 | + | ||
| 151 | + if save_model: | ||
| 152 | + # 保存模型 | ||
| 153 | + model_data = { | ||
| 154 | + 'encodings': encodings, | ||
| 155 | + 'names': names, | ||
| 156 | + 'classifier': self.face_classifier, | ||
| 157 | + 'label_encoder': self.label_encoder | ||
| 158 | + } | ||
| 159 | + | ||
| 160 | + model_path = os.path.join(self.model_save_path, 'face_recognizer.pkl') | ||
| 161 | + with open(model_path, 'wb') as f: | ||
| 162 | + pickle.dump(model_data, f) | ||
| 163 | + | ||
| 164 | + print(f"[INFO] 模型已保存到 {model_path}") | ||
| 165 | + | ||
| 166 | + print("[INFO] 分类器训练完成") | ||
| 167 | + | ||
| 168 | + def load_model(self, model_path='models/face_recognizer.pkl'): | ||
| 169 | + if not os.path.exists(model_path): | ||
| 170 | + print(f"[ERROR] 模型文件不存在: {model_path}") | ||
| 171 | + return False | ||
| 172 | + | ||
| 173 | + print("[INFO] 加载模型...") | ||
| 174 | + with open(model_path, 'rb') as f: | ||
| 175 | + model_data = pickle.load(f) | ||
| 176 | + | ||
| 177 | + self.known_face_encodings = model_data['encodings'] | ||
| 178 | + self.known_face_names = model_data['names'] | ||
| 179 | + self.face_classifier = model_data['classifier'] | ||
| 180 | + self.label_encoder = model_data['label_encoder'] | ||
| 181 | + | ||
| 182 | + print(f"[INFO] 模型加载成功，包含 {len(self.known_face_names)} 个人的数据") | ||
| 183 | + return True | ||
| 184 | + | ||
| 185 | + def recognize_faces(self, frame, draw_landmarks=False): | ||
| 186 | + if len(self.known_face_encodings) == 0: | ||
| 187 | + return frame, [] | ||
| 188 | + | ||
| 189 | + # 调整图片大小以提高处理速度 | ||
| 190 | + small_frame = cv2.resize(frame, (0, 0), fx=0.5, fy=0.5) | ||
| 191 | + rgb_small_frame = cv2.cvtColor(small_frame, cv2.COLOR_BGR2RGB) | ||
| 192 | + | ||
| 193 | + # 检测人脸位置 | ||
| 194 | + face_locations = face_recognition.face_locations( | ||
| 195 | + rgb_small_frame, model='hog' | ||
| 196 | + ) | ||
| 197 | + | ||
| 198 | + # 提取人脸编码 | ||
| 199 | + face_encodings = face_recognition.face_encodings( | ||
| 200 | + rgb_small_frame, face_locations | ||
| 201 | + ) | ||
| 202 | + | ||
| 203 | + face_names = [] | ||
| 204 | + face_boxes = [] | ||
| 205 | + | ||
| 206 | + for face_encoding, face_location in zip(face_encodings, face_locations): | ||
| 207 | + # 比较人脸 | ||
| 208 | + matches = face_recognition.compare_faces( | ||
| 209 | + self.known_face_encodings, | ||
| 210 | + face_encoding, | ||
| 211 | + tolerance=self.tolerance | ||
| 212 | + ) | ||
| 213 | + | ||
| 214 | + # 计算距离 | ||
| 215 | + face_distances = face_recognition.face_distance( | ||
| 216 | + self.known_face_encodings, | ||
| 217 | + face_encoding | ||
| 218 | + ) | ||
| 219 | + | ||
| 220 | + # 找到最佳匹配 | ||
| 221 | + best_match_index = np.argmin(face_distances) | ||
| 222 | + | ||
| 223 | + if matches[best_match_index]: | ||
| 224 | + name = self.known_face_names[best_match_index] | ||
| 225 | + distance = face_distances[best_match_index] | ||
| 226 | + | ||
| 227 | + if distance < self.tolerance: | ||
| 228 | + label = f"{name} ({distance:.2f})" | ||
| 229 | + else: | ||
| 230 | + label = "Unknown" | ||
| 231 | + else: | ||
| 232 | + label = "Unknown" | ||
| 233 | + | ||
| 234 | + # 调整坐标（因为之前缩小了图片） | ||
| 235 | + (top, right, bottom, left) = face_location | ||
| 236 | + top *= 2; right *= 2; bottom *= 2; left *= 2 | ||
| 237 | + | ||
| 238 | + face_names.append(label) | ||
| 239 | + face_boxes.append((left, top, right, bottom)) | ||
| 240 | + | ||
| 241 | + return self.draw_face_info(frame, face_boxes, face_names, draw_landmarks) | ||
| 242 | + | ||
| 243 | + def draw_face_info(self, frame, face_boxes, face_names, draw_landmarks=False): | ||
| 244 | + for (left, top, right, bottom), name in zip(face_boxes, face_names): | ||
| 245 | + # 绘制人脸框 | ||
| 246 | + color = (0, 255, 0) if "Unknown" not in name else (0, 0, 255) | ||
| 247 | + cv2.rectangle(frame, (left, top), (right, bottom), color, 2) | ||
| 248 | + | ||
| 249 | + # 绘制姓名标签 | ||
| 250 | + label_bg = (left, top - 35, right - left, 35) | ||
| 251 | + cv2.rectangle(frame, | ||
| 252 | + (label_bg[0], label_bg[1]), | ||
| 253 | + (label_bg[0] + label_bg[2], label_bg[1] + label_bg[3]), | ||
| 254 | + color, -1) | ||
| 255 | + | ||
| 256 | + cv2.putText(frame, name, (left + 6, top - 6), | ||
| 257 | + cv2.FONT_HERSHEY_DUPLEX, 0.8, (255, 255, 255), 1) | ||
| 258 | + | ||
| 259 | + if draw_landmarks: | ||
| 260 | + # 绘制面部特征点 | ||
| 261 | + face_image = frame[top:bottom, left:right] | ||
| 262 | + if face_image.size > 0: | ||
| 263 | + rgb_face = cv2.cvtColor(face_image, cv2.COLOR_BGR2RGB) | ||
| 264 | + landmarks = face_recognition.face_landmarks(rgb_face) | ||
| 265 | + | ||
| 266 | + for face_landmarks in landmarks: | ||
| 267 | + for facial_feature in face_landmarks.values(): | ||
| 268 | + for point in facial_feature: | ||
| 269 | + px, py = point | ||
| 270 | + cv2.circle(face_image, (px, py), 2, (255, 0, 0), -1) | ||
| 271 | + | ||
| 272 | + return frame, list(zip(face_names, face_boxes)) | ||
| 273 | + | ||
| 274 | + def realtime_recognition(self, show_fps=True, record_video=False): | ||
| 275 | + print("[INFO] 启动实时人脸识别...") | ||
| 276 | + print("[INFO] 按 'q' 键退出") | ||
| 277 | + print("[INFO] 按 's' 键截图") | ||
| 278 | + | ||
| 279 | + cap = cv2.VideoCapture(0) | ||
| 280 | + if not cap.isOpened(): | ||
| 281 | + print("[ERROR] 无法打开摄像头") | ||
| 282 | + return | ||
| 283 | + | ||
| 284 | + # 设置视频参数 | ||
| 285 | + cap.set(cv2.CAP_PROP_FRAME_WIDTH, 640) | ||
| 286 | + cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 480) | ||
| 287 | + | ||
| 288 | + # 视频录制设置 | ||
| 289 | + writer = None | ||
| 290 | + if record_video: | ||
| 291 | + fourcc = cv2.VideoWriter_fourcc(*'XVID') | ||
| 292 | + timestamp = time.strftime("%Y%m%d_%H%M%S") | ||
| 293 | + output_path = f"output/recording_{timestamp}.avi" | ||
| 294 | + writer = cv2.VideoWriter(output_path, fourcc, 20.0, (640, 480)) | ||
| 295 | + | ||
| 296 | + fps_time = time.time() | ||
| 297 | + frame_count = 0 | ||
| 298 | + fps = 0 | ||
| 299 | + | ||
| 300 | + while True: | ||
| 301 | + ret, frame = cap.read() | ||
| 302 | + if not ret: | ||
| 303 | + break | ||
| 304 | + | ||
| 305 | + # 人脸识别 | ||
| 306 | + processed_frame, results = self.recognize_faces(frame) | ||
| 307 | + | ||
| 308 | + # 计算FPS | ||
| 309 | + frame_count += 1 | ||
| 310 | + if time.time() - fps_time >= 1.0: | ||
| 311 | + fps = frame_count | ||
| 312 | + frame_count = 0 | ||
| 313 | + fps_time = time.time() | ||
| 314 | + | ||
| 315 | + if show_fps: | ||
| 316 | + cv2.putText(processed_frame, f"FPS: {fps}", | ||
| 317 | + (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 255), 2) | ||
| 318 | + | ||
| 319 | + # 显示检测到的人数 | ||
| 320 | + cv2.putText(processed_frame, f"人数: {len(results)}", | ||
| 321 | + (10, 60), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 255), 2) | ||
| 322 | + | ||
| 323 | + # 显示视频 | ||
| 324 | + cv2.imshow('人脸识别系统 - 按q退出', processed_frame) | ||
| 325 | + | ||
| 326 | + # 录制视频 | ||
| 327 | + if writer is not None: | ||
| 328 | + writer.write(processed_frame) | ||
| 329 | + | ||
| 330 | + # 键盘控制 | ||
| 331 | + key = cv2.waitKey(1) & 0xFF | ||
| 332 | + if key == ord('q'): | ||
| 333 | + break | ||
| 334 | + elif key == ord('s'): | ||
| 335 | + # 截图保存 | ||
| 336 | + timestamp = time.strftime("%Y%m%d_%H%M%S") | ||
| 337 | + screenshot_path = f"output/screenshot_{timestamp}.jpg" | ||
| 338 | + cv2.imwrite(screenshot_path, processed_frame) | ||
| 339 | + print(f"[INFO] 截图已保存: {screenshot_path}") | ||
| 340 | + | ||
| 341 | + # 清理资源 | ||
| 342 | + cap.release() | ||
| 343 | + if writer is not None: | ||
| 344 | + writer.release() | ||
| 345 | + cv2.destroyAllWindows() | ||
| 346 | + print("[INFO] 实时识别已停止") | ||
| 347 | + | ||
| 348 | +class AdvancedFaceAnalyzer: | ||
| 349 | + | ||
| 350 | + def __init__(self): | ||
| 351 | + self.detector = dlib.get_frontal_face_detector() | ||
| 352 | + self.predictor = dlib.shape_predictor( | ||
| 353 | + 'shape_predictor_68_face_landmarks.dat' | ||
| 354 | + ) | ||
| 355 | + | ||
| 356 | + def download_landmark_model(self): | ||
| 357 | + import urllib.request | ||
| 358 | + model_url = "http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2" | ||
| 359 | + model_path = 'shape_predictor_68_face_landmarks.dat.bz2' | ||
| 360 | + | ||
| 361 | + if not os.path.exists('shape_predictor_68_face_landmarks.dat'): | ||
| 362 | + print("[INFO] 下载面部特征点模型...") | ||
| 363 | + urllib.request.urlretrieve(model_url, model_path) | ||
| 364 | + | ||
| 365 | + import bz2 | ||
| 366 | + with bz2.BZ2File(model_path, 'rb') as fr, \ | ||
| 367 | + open('shape_predictor_68_face_landmarks.dat', 'wb') as fw: | ||
| 368 | + fw.write(fr.read()) | ||
| 369 | + | ||
| 370 | + os.remove(model_path) | ||
| 371 | + print("[INFO] 模型下载完成") | ||
| 372 | + | ||
| 373 | + def analyze_face_attributes(self, image): | ||
| 374 | + gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) | ||
| 375 | + faces = self.detector(gray) | ||
| 376 | + | ||
| 377 | + results = [] | ||
| 378 | + for face in faces: | ||
| 379 | + landmarks = self.predictor(gray, face) | ||
| 380 | + | ||
| 381 | + # 计算各种特征 | ||
| 382 | + attributes = { | ||
| 383 | + 'face_rect': (face.left(), face.top(), face.right(), face.bottom()), | ||
| 384 | + 'eye_aspect_ratio': self.eye_aspect_ratio(landmarks), | ||
| 385 | + 'mouth_aspect_ratio': self.mouth_aspect_ratio(landmarks), | ||
| 386 | + 'head_pose': self.estimate_head_pose(landmarks) | ||
| 387 | + } | ||
| 388 | + results.append(attributes) | ||
| 389 | + | ||
| 390 | + return results | ||
| 391 | + | ||
| 392 | + def eye_aspect_ratio(self, landmarks): | ||
| 393 | + # 左眼 | ||
| 394 | + left_eye = landmarks.part(36), landmarks.part(37), landmarks.part(38), \ | ||
| 395 | + landmarks.part(39), landmarks.part(40), landmarks.part(41) | ||
| 396 | + | ||
| 397 | + # 右眼 | ||
| 398 | + right_eye = landmarks.part(42), landmarks.part(43), landmarks.part(44), \ | ||
| 399 | + landmarks.part(45), landmarks.part(46), landmarks.part(47) | ||
| 400 | + | ||
| 401 | + def eye_ratio(eye): | ||
| 402 | + # 计算垂直距离 | ||
| 403 | + A = np.linalg.norm(np.array([eye[1].x, eye[1].y]) - | ||
| 404 | + np.array([eye[5].x, eye[5].y])) | ||
| 405 | + B = np.linalg.norm(np.array([eye[2].x, eye[2].y]) - | ||
| 406 | + np.array([eye[4].x, eye[4].y])) | ||
| 407 | + | ||
| 408 | + # 计算水平距离 | ||
| 409 | + C = np.linalg.norm(np.array([eye[0].x, eye[0].y]) - | ||
| 410 | + np.array([eye[3].x, eye[3].y])) | ||
| 411 | + | ||
| 412 | + return (A + B) / (2.0 * C) | ||
| 413 | + | ||
| 414 | + return (eye_ratio(left_eye) + eye_ratio(right_eye)) / 2.0 | ||
| 415 | + | ||
| 416 | + def mouth_aspect_ratio(self, landmarks): | ||
| 417 | + """计算嘴巴纵横比（用于检测是否张嘴）""" | ||
| 418 | + mouth_points = [ | ||
| 419 | + landmarks.part(i) for i in range(48, 68) | ||
| 420 | + ] | ||
| 421 | + | ||
| 422 | + # 计算垂直距离 | ||
| 423 | + vertical = np.linalg.norm( | ||
| 424 | + np.array([mouth_points[2].x, mouth_points[2].y]) - | ||
| 425 | + np.array([mouth_points[10].x, mouth_points[10].y]) | ||
| 426 | + ) | ||
| 427 | + | ||
| 428 | + # 计算水平距离 | ||
| 429 | + horizontal = np.linalg.norm( | ||
| 430 | + np.array([mouth_points[0].x, mouth_points[0].y]) - | ||
| 431 | + np.array([mouth_points[6].x, mouth_points[6].y]) | ||
| 432 | + ) | ||
| 433 | + | ||
| 434 | + return vertical / horizontal | ||
| 435 | + | ||
| 436 | + def estimate_head_pose(self, landmarks): | ||
| 437 | + """估计头部姿态""" | ||
| 438 | + # 简化的头部姿态估计 | ||
| 439 | + image_points = np.array([ | ||
| 440 | + [landmarks.part(30).x, landmarks.part(30).y], # 鼻尖 | ||
| 441 | + [landmarks.part(8).x, landmarks.part(8).y], # 下巴 | ||
| 442 | + [landmarks.part(36).x, landmarks.part(36).y], # 左眼眼角 | ||
| 443 | + [landmarks.part(45).x, landmarks.part(45).y], # 右眼眼角 | ||
| 444 | + [landmarks.part(48).x, landmarks.part(48).y], # 左嘴角 | ||
| 445 | + [landmarks.part(54).x, landmarks.part(54).y] # 右嘴角 | ||
| 446 | + ], dtype="double") | ||
| 447 | + | ||
| 448 | + # 3D模型点 | ||
| 449 | + model_points = np.array([ | ||
| 450 | + (0.0, 0.0, 0.0), # 鼻尖 | ||
| 451 | + (0.0, -330.0, -65.0), # 下巴 | ||
| 452 | + (-165.0, 170.0, -135.0), # 左眼眼角 | ||
| 453 | + (165.0, 170.0, -135.0), # 右眼眼角 | ||
| 454 | + ]) | ||
| 455 | + | ||
| 456 | + # 这里简化处理，实际需要相机参数 | ||
| 457 | + return "head_pose_estimation_placeholder" | ||
| 458 | + | ||
| 459 | +def main(): | ||
| 460 | + """主函数""" | ||
| 461 | + print("=" * 50) | ||
| 462 | + print("人脸识别系统") | ||
| 463 | + print("=" * 50) | ||
| 464 | + | ||
| 465 | + # 创建识别器 | ||
| 466 | + recognizer = FaceRecognizer(method='hog', tolerance=0.6) | ||
| 467 | + | ||
| 468 | + while True: | ||
| 469 | + print("\n请选择操作:") | ||
| 470 | + print("1. 创建新的人脸数据集") | ||
| 471 | + print("2. 训练人脸识别模型") | ||
| 472 | + print("3. 实时人脸识别") | ||
| 473 | + print("4. 加载已有模型并识别") | ||
| 474 | + print("5. 退出") | ||
| 475 | + | ||
| 476 | + choice = input("请输入选项 (1-5): ").strip() | ||
| 477 | + | ||
| 478 | + if choice == '1': | ||
| 479 | + name = input("请输入姓名: ").strip() | ||
| 480 | + if name: | ||
| 481 | + recognizer.create_dataset(name, num_samples=20) | ||
| 482 | + | ||
| 483 | + elif choice == '2': | ||
| 484 | + recognizer.train_classifier(save_model=True) | ||
| 485 | + | ||
| 486 | + elif choice == '3': | ||
| 487 | + if not recognizer.load_model(): | ||
| 488 | + print("[INFO] 未找到模型，请先训练模型") | ||
| 489 | + continue | ||
| 490 | + | ||
| 491 | + show_fps = input("显示FPS? (y/n): ").lower() == 'y' | ||
| 492 | + record = input("录制视频? (y/n): ").lower() == 'y' | ||
| 493 | + recognizer.realtime_recognition(show_fps=show_fps, record_video=record) | ||
| 494 | + | ||
| 495 | + elif choice == '4': | ||
| 496 | + if recognizer.load_model(): | ||
| 497 | + show_fps = input("显示FPS? (y/n): ").lower() == 'y' | ||
| 498 | + record = input("录制视频? (y/n): ").lower() == 'y' | ||
| 499 | + recognizer.realtime_recognition(show_fps=show_fps, record_video=record) | ||
| 500 | + | ||
| 501 | + elif choice == '5': | ||
| 502 | + print("[INFO] 退出系统") | ||
| 503 | + break | ||
| 504 | + | ||
| 505 | + else: | ||
| 506 | + print("[ERROR] 无效的选项，请重新输入") | ||
| 507 | + | ||
| 508 | +if __name__ == "__main__": | ||
| 509 | + | ||
| 510 | + main() | ||