def face_detect(input_img, pnet, rnet, onet,minsize=60, threshold=[0.6, 0.7, 0.7], factor=0.707): frame_to_face = input_img img_size = frame_to_face.shape frame_to_face = frame_to_face[..., ::-1] bounding_boxes, _ = FaceRecognition.detect_face.detect_face(frame_to_face, minsize, pnet, rnet, onet, threshold, factor) if len(bounding_boxes) > 0: faces = [] faces_boxes = [] faces_confidence = [] for i in range(len(bounding_boxes)): det = np.squeeze(bounding_boxes[i, 0:4]) bb = np.zeros(4, dtype=np.int32) margin_x = (det[2] - det[0]) / 4 margin_y = (det[3] - det[1]) / 4 assert margin_x > 0 and margin_y > 0 bb[0] = np.maximum(det[0] - margin_x, 0) bb[1] = np.maximum(det[1] - margin_y, 0) bb[2] = np.minimum(det[2] + margin_x, img_size[1]) bb[3] = np.minimum(det[3] + margin_y / 2, img_size[0]) # output_image = cv2.rectangle(output_image, (bb[0], bb[1]), (bb[2], bb[3]), (0, 255, 0), 3) # bounding_boxes = find_face_from_openpose(keypoints) # for bb in bounding_boxes: face_img = frame_to_face[bb[1]:bb[3], bb[0]:bb[2], :] face_img = cv2.resize(face_img, (160, 160)) face_img = facenet.prewhiten(face_img) faces.append(face_img) faces_boxes.append(bb) else: faces = [] faces_boxes = [] return faces, faces_boxes
def face_detect_and_recognize(input_img, sess, pnet, rnet, onet, images_placeholder, phase_train_placeholder, embeddings, minsize, threshold, factor): frame_to_face = input_img img_size = frame_to_face.shape frame_to_face = frame_to_face[..., ::-1] bounding_boxes, _ = FaceRecognition.detect_face.detect_face(frame_to_face, minsize, pnet, rnet, onet, threshold, factor) if len(bounding_boxes) > 0: faces = [] faces_boxes = [] faces_confidence = [] for i in range(len(bounding_boxes)): det = np.squeeze(bounding_boxes[i, 0:4]) bb = np.zeros(4, dtype=np.int32) margin_x = (det[2] - det[0]) / 4 margin_y = (det[3] - det[1]) / 4 assert margin_x > 0 and margin_y > 0 bb[0] = np.maximum(det[0] - margin_x, 0) bb[1] = np.maximum(det[1] - margin_y, 0) bb[2] = np.minimum(det[2] + margin_x, img_size[1]) bb[3] = np.minimum(det[3] + margin_y, img_size[0]) # output_image = cv2.rectangle(output_image, (bb[0], bb[1]), (bb[2], bb[3]), (0, 255, 0), 3) # bounding_boxes = find_face_from_openpose(keypoints) # for bb in bounding_boxes: face_img = frame_to_face[bb[1]:bb[3], bb[0]:bb[2], :] face_img = cv2.resize(face_img, (160, 160)) face_img = facenet.prewhiten(face_img) faces.append(face_img) faces_boxes.append(bb) feed_dict = {images_placeholder: faces, phase_train_placeholder: False} emb = sess.run(embeddings, feed_dict=feed_dict) else: faces_boxes = [] emb = [] return faces_boxes, emb
def resultado(self): #input_video="http://*****:*****@192.168.1.33:8081" #input_video="./FaceRecognition/videoGOT.mp4" #ruta modeldir = os.path.realpath( "./FaceRecognition/model/20170511-185253.pb" ) # 'C:/Users/Corgito/Desktop/untitled/FaceRecognition/model/20170511-185253.pb' classifier_filename = os.path.realpath( "./FaceRecognition/class/classifier.pkl" ) # 'C:/Users/Corgito/Desktop/untitled/FaceRecognition/class/classifier.pkl' npy = os.path.realpath( "./FaceRecognition/npy" ) # 'C:/Users/Corgito/Desktop/untitled/FaceRecognition/npy' train_img = os.path.realpath( "./FaceRecognition/train_img" ) # "C:/Users/Corgito/Desktop/untitled/FaceRecognition/train_img" with tf.Graph().as_default(): gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.6) sess = tf.Session(config=tf.ConfigProto( gpu_options=gpu_options, log_device_placement=False)) with sess.as_default(): pnet, rnet, onet = detect_face.create_mtcnn(sess, npy) minsize = 10 # minimum size of face threshold = [0.8, 0.8, 0.8] # three steps's threshold factor = 0.709 # scale factor margin = 44 frame_interval = 3 batch_size = 1000 image_size = 182 input_image_size = 160 HumanNames = os.listdir(train_img) HumanNames.sort() print('Loading Modal') facenet.load_model(modeldir) images_placeholder = tf.get_default_graph().get_tensor_by_name( "input:0") embeddings = tf.get_default_graph().get_tensor_by_name( "embeddings:0") phase_train_placeholder = tf.get_default_graph( ).get_tensor_by_name("phase_train:0") embedding_size = embeddings.get_shape()[1] classifier_filename_exp = os.path.expanduser( classifier_filename) with open(classifier_filename_exp, 'rb') as infile: (model, class_names) = pickle.load(infile) video_capture = cv2.VideoCapture(0) c = 0 print('Start Recognition') prevTime = 0 while True: ret, frame = video_capture.read() #frame = cv2.resize(frame, (0, 0), fx=0.5, fy=0.5) # resize frame (optional) curTime = time.time() + 1 # calc fps timeF = frame_interval if (c % timeF == 0): find_results = [] if frame.ndim == 2: frame = facenet.to_rgb(frame) frame = frame[:, :, 0:3] bounding_boxes, _ = detect_face.detect_face( frame, minsize, pnet, rnet, onet, threshold, factor) nrof_faces = bounding_boxes.shape[0] print('Detected_FaceNum: %d' % nrof_faces) if nrof_faces > 0: det = bounding_boxes[:, 0:4] img_size = np.asarray(frame.shape)[0:2] cropped = [] scaled = [] scaled_reshape = [] bb = np.zeros((nrof_faces, 4), dtype=np.int32) for i in range(nrof_faces): emb_array = np.zeros((1, embedding_size)) bb[i][0] = det[i][0] bb[i][1] = det[i][1] bb[i][2] = det[i][2] bb[i][3] = det[i][3] # inner exception if bb[i][0] <= 0 or bb[i][1] <= 0 or bb[i][ 2] >= len(frame[0]) or bb[i][3] >= len( frame): print('Face is very close!') continue cropped.append(frame[bb[i][1]:bb[i][3], bb[i][0]:bb[i][2], :]) cropped[i] = facenet.flip(cropped[i], False) scaled.append( misc.imresize(cropped[i], (image_size, image_size), interp='bilinear')) scaled[i] = cv2.resize( scaled[i], (input_image_size, input_image_size), interpolation=cv2.INTER_CUBIC) scaled[i] = facenet.prewhiten(scaled[i]) scaled_reshape.append(scaled[i].reshape( -1, input_image_size, input_image_size, 3)) feed_dict = { images_placeholder: scaled_reshape[i], phase_train_placeholder: False } emb_array[0, :] = sess.run(embeddings, feed_dict=feed_dict) predictions = model.predict_proba(emb_array) print(predictions) best_class_indices = np.argmax(predictions, axis=1) best_class_probabilities = predictions[ np.arange(len(best_class_indices)), best_class_indices] # print("predictions") print(best_class_indices, ' with accuracy ', best_class_probabilities) # print(best_class_probabilities) if best_class_probabilities > 0.1: cv2.rectangle(frame, (bb[i][0], bb[i][1]), (bb[i][2], bb[i][3]), (0, 255, 0), 2) # boxing face # plot result idx under box text_x = bb[i][0] text_y = bb[i][3] + 20 print('Result Indices: ', best_class_indices[0]) print(HumanNames) for H_i in HumanNames: if HumanNames[ best_class_indices[0]] == H_i: result_names = HumanNames[ best_class_indices[0]] cv2.putText( frame, result_names, (text_x, text_y), cv2.FONT_HERSHEY_COMPLEX_SMALL, 1, (0, 0, 255), thickness=1, lineType=2) else: print('Alignment Failure') # c+=1 #cv2.imshow('Video', frame) evento, cuadro = cv2.imencode('.jpg', frame) #ventana='<video width="320" height="240" controls> <source src="',cuadro.tobytes(),'" type="video/mp4"> </video>' ventana = cuadro.tobytes() yield (b'--frame\r\n' b'Content-Type: video/mp4\r\n\r\n' + ventana + b'\r\n\r\n') if cv2.waitKey(32) & 0xFF == ord('q'): break video_capture.release() cv2.destroyAllWindows()
def resultado(self, ruta): """ Método para aplicar reconocimiento facial a la imagen. :Método función: **resultado** (self,input_video). :valor self: None. :valor input_video: Ruta del archivo de imagen. :Lista de parámetros variables: :minsize: Tamaño mínimo para detectar el rostro. :umbral: Valor mínimo para marcar a la persona como desconocida. :return result_name: Devuelve el nombre de la persona reconocida. """ result_names = "No se detecto ninguna cara" img_path = ruta modeldir = os.path.realpath( "./FaceRecognition/model/20170511-185253.pb") classifier_filename = os.path.realpath( "./FaceRecognition/class/classifier.pkl") npy = os.path.realpath("./FaceRecognition/npy") train_img = os.path.realpath("./FaceRecognition/train_img") with tf.Graph().as_default(): gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.6) sess = tf.Session(config=tf.ConfigProto( gpu_options=gpu_options, log_device_placement=False)) with sess.as_default(): pnet, rnet, onet = detect_face.create_mtcnn(sess, npy) image = cv2.imread(img_path) umbral = 0.35 h, w, _ = image.shape minsize = 8 # minimum size of face threshold = [0.7, 0.7, 0.7] # three steps's threshold factor = 0.709 # scale factor margin = 20 frame_interval = 3 batch_size = 1000 image_size = 182 input_image_size = 160 HumanNames = os.listdir(train_img) HumanNames.sort() print('Loading feature extraction model') facenet.load_model(modeldir) images_placeholder = tf.get_default_graph().get_tensor_by_name( "input:0") embeddings = tf.get_default_graph().get_tensor_by_name( "embeddings:0") phase_train_placeholder = tf.get_default_graph( ).get_tensor_by_name("phase_train:0") embedding_size = embeddings.get_shape()[1] classifier_filename_exp = os.path.expanduser( classifier_filename) with open(classifier_filename_exp, 'rb') as infile: (model, class_names) = pickle.load(infile) c = 0 print('Start Recognition!') prevTime = 0 frame = cv2.imread(img_path, 0) curTime = time.time() + 1 # calc fps timeF = frame_interval if (c % timeF == 0): find_results = [] if frame.ndim == 2: frame = facenet.to_rgb(frame) frame = frame[:, :, 0:3] bounding_boxes, _ = detect_face.detect_face( frame, minsize, pnet, rnet, onet, threshold, factor) nrof_faces = bounding_boxes.shape[0] print('Face Detected: %d' % nrof_faces) if nrof_faces > 0: det = bounding_boxes[:, 0:4] img_size = np.asarray(frame.shape)[0:2] cropped = [] scaled = [] scaled_reshape = [] bb = np.zeros((nrof_faces, 4), dtype=np.int32) for i in range(nrof_faces): emb_array = np.zeros((1, embedding_size)) bb[i][0] = det[i][0] bb[i][1] = det[i][1] bb[i][2] = det[i][2] bb[i][3] = det[i][3] # inner exception if bb[i][0] <= 0 or bb[i][1] <= 0 or bb[i][ 2] >= len( frame[0]) or bb[i][3] >= len(frame): print('face is too close') continue cropped.append(frame[bb[i][1]:bb[i][3], bb[i][0]:bb[i][2], :]) cropped[i] = facenet.flip(cropped[i], False) scaled.append( misc.imresize(cropped[i], (image_size, image_size), interp='bilinear')) scaled[i] = cv2.resize( scaled[i], (input_image_size, input_image_size), interpolation=cv2.INTER_CUBIC) scaled[i] = facenet.prewhiten(scaled[i]) scaled_reshape.append(scaled[i].reshape( -1, input_image_size, input_image_size, 3)) feed_dict = { images_placeholder: scaled_reshape[i], phase_train_placeholder: False } emb_array[0, :] = sess.run(embeddings, feed_dict=feed_dict) predictions = model.predict_proba(emb_array) print(predictions) best_class_indices = np.argmax(predictions, axis=1) # print(best_class_indices) best_class_probabilities = predictions[ np.arange(len(best_class_indices)), best_class_indices] print(best_class_probabilities) cv2.rectangle(frame, (bb[i][0], bb[i][1]), (bb[i][2], bb[i][3]), (0, 255, 0), 2) # boxing face # plot result idx under box text_x = bb[i][0] text_y = bb[i][3] + 20 print('Result Indices: ', best_class_indices[0]) print(HumanNames) print("Mejor Indicie", best_class_indices[0]) if best_class_probabilities < umbral: result_names = "Desconocido" cv2.putText(frame, result_names + str(i), (text_x, text_y), cv2.FONT_HERSHEY_COMPLEX_SMALL, 1, (0, 0, 255), thickness=1, lineType=2) else: for H_i in HumanNames: if HumanNames[ best_class_indices[0]] == H_i: result_names = HumanNames[ best_class_indices[0]] # + str(i) cv2.putText( frame, result_names, (text_x, text_y), cv2.FONT_HERSHEY_COMPLEX_SMALL, 1, (0, 0, 255), thickness=1, lineType=2) print("La cara", i, "pertene a", result_names, "con un indice de probabilidad de", best_class_probabilities) else: print('Unable to align') #dim=int(frame.shape[1]-200),int(frame.shape[0]-200)#height,width #frame = cv2.resize(frame, dim, fx=0.5, fy=0.5) #el resize es opcional, pero si se hace mejor despues de la deteccíon ya que si no perdemos información #cv2.imshow('Image', frame) cv2.imwrite(ruta, frame) return result_names if cv2.waitKey(1000000) & 0xFF == ord('q'): cv2.destroyAllWindows() #sys.exit("Thanks") cv2.destroyAllWindows()
def process_latest_frame(openpose_path, mtcnn_path, facenet_path, reid_path, queue=None): frame_count = 1 # Load OpenPose Model openpose = load_openpose_model(openpose_path) print('----------OpenPose Loaded----------') # Load MTCNN (face detection) minsize = 50 # minimum size of face threshold = [0.6, 0.7, 0.7] # three steps's threshold factor = 0.709 margin = 44 with tf.Graph().as_default(): config = tf.ConfigProto() config.gpu_options.allow_growth = True sess1 = tf.Session(config=config) with sess1.as_default(): pnet, rnet, onet = FaceRecognition.detect_face.create_mtcnn( sess1, mtcnn_path) print('----------MTCNN Loaded----------') # Load Facenet config = tf.ConfigProto() config.gpu_options.allow_growth = True sess = tf.Session(config=config) facenet.load_model(facenet_path) images_placeholder = tf.get_default_graph().get_tensor_by_name("input:0") embeddings = tf.get_default_graph().get_tensor_by_name("embeddings:0") phase_train_placeholder = tf.get_default_graph().get_tensor_by_name( "phase_train:0") print('----------FaceNet Loaded----------') # Load Reid model reid_model = load_reid_model(reid_path) print('----------ReID Loaded----------') while True: if not queue.empty(): frame_count += 1 start = time.time() frame_to_process = queue.get() img_size = np.asarray(frame_to_process.shape)[0:2] keypoints, output_image = openpose.forward(frame_to_process, True) if frame_count % 200 == 0: print('--------------Do face recognition-------------') frame_to_face = frame_to_process.copy() frame_to_face = frame_to_face[..., ::-1] bounding_boxes, _ = FaceRecognition.detect_face.detect_face( frame_to_face, minsize, pnet, rnet, onet, threshold, factor) if len(bounding_boxes) > 0: det = np.squeeze(bounding_boxes[0, 0:4]) bb = np.zeros(4, dtype=np.int32) bb[0] = np.maximum(det[0] - margin / 2, 0) bb[1] = np.maximum(det[1] - margin / 2, 0) bb[2] = np.minimum(det[2] + margin / 2, img_size[1]) bb[3] = np.minimum(det[3] + margin / 2, img_size[0]) # output_image = cv2.rectangle(output_image, (bb[0], bb[1]), (bb[2], bb[3]), (0, 255, 0), 3) # bounding_boxes = find_face_from_openpose(keypoints) # for bb in bounding_boxes: face_img = frame_to_face[bb[1]:bb[3], bb[0]:bb[2], :] face_img = cv2.resize(face_img, (160, 160)) face_img = facenet.prewhiten(face_img) feed_dict = { images_placeholder: [face_img], phase_train_placeholder: False } emb = sess.run(embeddings, feed_dict=feed_dict) output_image = cv2.rectangle(output_image, (bb[0], bb[1]), (bb[2], bb[3]), (0, 255, 0), 3) print('Frame: {:>7}, {:4}'.format(frame_count, time.time() - start)) cv2.imshow('a', output_image) cv2.waitKey(1)