def main(args): with open(args.pickle,'rb') as f: feature_array = pickle.load(f, encoding='utf-8') #model_exp = '../lib/src/ckpt/20170512-110547' model_exp = args.model graph_fr = tf.Graph() sess_fr = tf.Session(graph=graph_fr) with graph_fr.as_default(): with sess_fr.as_default(): #saverf = tf.train.import_meta_graph(os.path.join(model_exp, 'model-20180408-102900.meta')) #saverf.restore(sess_fr, os.path.join(model_exp, 'model-20180408-102900.ckpt-90')) load_model(model_exp) pnet, rnet, onet = detect_face.create_mtcnn(sess_fr, None) retrieve.recognize_face(sess_fr,pnet, rnet, onet,feature_array)
return render_template("index.html") @app.route("/predict") def predict_page(): """Renders the 'predict.html' page for manual image file uploads for prediction.""" return render_template("predict.html") if __name__ == '__main__': """Server and FaceNet Tensorflow configuration.""" # Load FaceNet model and configure placeholders for forward pass into the FaceNet model to calculate embeddings model_path = 'model/20170512-110547/20170512-110547.pb' facenet_model = load_model(model_path) config = tf.ConfigProto() config.gpu_options.allow_growth = True image_size = 160 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") # Initiate persistent FaceNet model in memory facenet_persistent_session = tf.Session(graph=facenet_model, config=config) # Create Multi-Task Cascading Convolutional (MTCNN) neural networks for Face Detection pnet, rnet, onet = detect_face.create_mtcnn(sess=facenet_persistent_session, model_path=None) # Start flask application on waitress WSGI server serve(app=app, host='0.0.0.0', port=5000)
# Loading the stored face embedding vectors for image retrieval # # #============================================================================================================================== with open('../lib/src/face_embeddings.pickle', 'rb') as f: feature_array = pickle.load(f) model_exp = '../lib/src/ckpt/20170512-110547' graph_fr = tf.Graph() sess_fr = tf.Session(graph=graph_fr) with graph_fr.as_default(): saverf = tf.train.import_meta_graph( os.path.join(model_exp, 'model-20170512-110547.meta')) saverf.restore( sess_fr, os.path.join(model_exp, 'model-20170512-110547.ckpt-250000')) pnet, rnet, onet = detect_face.create_mtcnn(sess_fr, None) #============================================================================================================================== # # This function is used to do the face recognition from video camera # # #============================================================================================================================== @app.route('/facerecognitionLive', methods=['GET', 'POST']) def face_det(): retrieve.recognize_face(sess_fr, pnet, rnet, onet, feature_array) #==============================================================================================================================
def main(input_dir, output_dir, image_size, margin, random_order, gpu_memory_fraction, detect_multiple_faces): sleep(random.random()) output_dir = os.path.expanduser(output_dir) if not os.path.exists(output_dir): os.makedirs(output_dir) # Store some git revision info in a text file in the log directory #src_path,_ = os.path.split(os.path.realpath(__file__)) src_path = input_dir facenet.store_revision_info(src_path, output_dir, ' '.join(sys.argv)) dataset = facenet.get_dataset(input_dir) print('Creating networks and loading parameters') with tf.Graph().as_default(): gpu_options = tf.GPUOptions( per_process_gpu_memory_fraction=gpu_memory_fraction) 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, None) minsize = 20 # minimum size of face threshold = [0.6, 0.7, 0.7] # three steps's threshold factor = 0.709 # scale factor # Add a random key to the filename to allow alignment using multiple processes random_key = np.random.randint(0, high=99999) #bounding_boxes_filename = os.path.join(output_dir, 'bounding_boxes_%05d.txt' % random_key) if 1: nrof_images_total = 0 nrof_successfully_aligned = 0 if random_order: random.shuffle(dataset) for cls in dataset: print("cls.name", cls.name) output_class_dir = os.path.join(output_dir, cls.name) if not os.path.exists(output_class_dir): os.makedirs(output_class_dir) if random_order: random.shuffle(cls.image_paths) for image_path in cls.image_paths: nrof_images_total += 1 filename = os.path.splitext(os.path.split(image_path)[1])[0] output_filename = os.path.join(output_class_dir, filename + '.png') print(image_path) if not os.path.exists(output_filename): try: img = misc.imread(image_path) except (IOError, ValueError, IndexError) as e: errorMessage = '{}: {}'.format(image_path, e) print(errorMessage) else: if img.ndim < 2: print('Unable to align "%s"' % image_path) #text_file.write('%s\n' % (output_filename)) continue if img.ndim == 2: img = facenet.to_rgb(img) img = img[:, :, 0:3] bounding_boxes, _ = detect_face.detect_face( img, minsize, pnet, rnet, onet, threshold, factor) nrof_faces = bounding_boxes.shape[0] if nrof_faces > 0: det = bounding_boxes[:, 0:4] det_arr = [] img_size = np.asarray(img.shape)[0:2] if nrof_faces > 1: if detect_multiple_faces: for i in range(nrof_faces): det_arr.append(np.squeeze(det[i])) else: bounding_box_size = ( det[:, 2] - det[:, 0]) * (det[:, 3] - det[:, 1]) img_center = img_size / 2 offsets = np.vstack([ (det[:, 0] + det[:, 2]) / 2 - img_center[1], (det[:, 1] + det[:, 3]) / 2 - img_center[0] ]) offset_dist_squared = np.sum( np.power(offsets, 2.0), 0) index = np.argmax( bounding_box_size - offset_dist_squared * 2.0 ) # some extra weight on the centering det_arr.append(det[index, :]) else: det_arr.append(np.squeeze(det)) for i, det in enumerate(det_arr): det = np.squeeze(det) 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]) cropped = img[bb[1]:bb[3], bb[0]:bb[2], :] scaled = misc.imresize( cropped, (image_size, image_size), interp='bilinear') nrof_successfully_aligned += 1 output_filename_n = "{}_{}.{}".format( output_filename.split('.')[0], i, output_filename.split('.')[-1]) misc.imsave(output_filename_n, scaled) #text_file.write('%s %d %d %d %d\n' % (output_filename_n, bb[0], bb[1], bb[2], bb[3])) else: print('Unable to align "%s"' % image_path) #text_file.write('%s\n' % (output_filename)) os.remove(os.path.join(settings.BASE_DIR, f'media/datab/revision_info.txt')) print('Total number of images: %d' % nrof_images_total) print('Number of successfully aligned images: %d' % nrof_successfully_aligned)
def recognize(argv): dt = datetime.datetime #detector = dlib.get_frontal_face_detector() #detector = cv2.CascadeClassifier('haarcascade_frontalface_default.xml') detector = cv2.CascadeClassifier('haarcascade_frontalface_alt.xml') use_haar = False #True use_dlib = False with open(argv.pickle, 'rb') as f: sys.stderr.write("will load feature\n") feature_array = pickle.load(f, encoding='utf-8') model_exp = argv.model graph_fr = tf.Graph() sess_fr = tf.Session(graph=graph_fr) with graph_fr.as_default(): with sess_fr.as_default(): image_size = (160, 160) sys.stderr.write("will load model\n") #sys.stderr.write("did load model\n") if not use_dlib and not use_haar: pnet, rnet, onet = detect_face.create_mtcnn(sess_fr, None) load_model(model_exp) #sys.stderr.write("will get placeholders\n") images_placeholder = sess_fr.graph.get_tensor_by_name( "input:0") images_placeholder = tf.image.resize_images( images_placeholder, image_size) embeddings = sess_fr.graph.get_tensor_by_name("embeddings:0") phase_train_placeholder = sess_fr.graph.get_tensor_by_name( "phase_train:0") downsample = shared['downsample'] shared['recog_ready'] = True while True: t = dt.utcnow() img = get_img() if img is not None: downsampleShape = (int(img.shape[1] / downsample), int(img.shape[0] / downsample)) #img = np.asarray(Image.fromarray(img).resize(downsampleShape, resample=Image.BILINEAR)) if downsample > 1: img = cv2.resize(img, downsampleShape) if use_dlib: result = retrieve.recognize_hog( images_placeholder, phase_train_placeholder, embeddings, sess_fr, feature_array, img, detector) elif use_haar: result = retrieve.recognize_haar( images_placeholder, phase_train_placeholder, embeddings, sess_fr, feature_array, img, detector) else: result = [] for roi in shared['candidate_area']: bbox = roi[0] cropped = roi[1] #print("align in {},{}: {},{}".format(bbox[0], bbox[1], bbox[2], bbox[3])) r = retrieve.recognize_mtcnn( images_placeholder, phase_train_placeholder, embeddings, sess_fr, pnet, rnet, onet, feature_array, cropped) for i, face in enumerate(r): b = face['box'] b[0] += bbox[0] b[1] += bbox[1] b[2] += bbox[0] b[3] += bbox[1] r[i]['box'] = b print("face: {},{},{},{}".format( b[0], b[1], b[2], b[3])) result.extend(r) update_result(result, t)