基于深度学习人脸静默活体人脸识别检测算法python源码+模型+操作说明.zip

import math import cv2 import time import numpy as np import os import keras from keras.layers import * import tensorflow as tf from mtcnn import MTCNN from keras.models import load_model model = load_model("../model/fas.h5") def load_mtcnn_model(model_path): mtcnn = MTCNN(model_path) return mtcnn def test_one(X): TEMP = X.copy() X = (cv2.resize(X,(224,224))-127.5)/127.5 t = model.predict(np.array([X]))[0] time_end=time.time() return t def test_camera(mtcnn,index = 0): cam=cv2.VideoCapture(0) while(1): success, frame = cam.read() if success: image = frame img_size = np.asarray(image.shape)[0:2] bounding_boxes, scores, landmarks = mtcnn.detect(image) nrof_faces = bounding_boxes.shape[0] if nrof_faces > 0: for det, pts in zip(bounding_boxes, landmarks): det = det.astype('int32') #print("face confidence: %2.3f" % confidence) det = np.squeeze(det) y1 = int(np.maximum(det[0], 0)) x1 = int(np.maximum(det[1], 0)) y2 = int(np.minimum(det[2], img_size[1]-1)) x2 = int(np.minimum(det[3], img_size[0]-1)) w = x2-x1 h = y2-y1 _r = int(max(w,h)*0.6) cx,cy = (x1+x2)//2, (y1+y2)//2 x1 = cx - _r y1 = cy - _r x1 = int(max(x1,0)) y1 = int(max(y1,0)) x2 = cx + _r y2 = cy + _r h,w,c =frame.shape x2 = int(min(x2 ,w-2)) y2 = int(min(y2, h-2)) _frame = frame[y1:y2 , x1:x2] score = test_one(_frame) print(score) if score > 0.95: cv2.rectangle(frame, (x1,y1) ,(x2,y2) , (0,255,0) ,2) else: cv2.rectangle(frame, (x1,y1) ,(x2,y2) , (255,0,0) ,2) cv2.imshow("image", image) if cv2.waitKey(1) & 0XFF == ord('q'): cv2.destroyAllWindows() cam.release() break else: cv2.destroyAllWindows() cam.release() break if __name__ == '__main__': mtcnn = load_mtcnn_model("../model/mtcnn.pb") test_camera(mtcnn)