def __init__(self, line, sline1, sline2): #line is for counting #slines for speed estimation self.deepsort=deepsort_rbc() self.yolov3 = YOLOV3("cfg/yolo_2k_reanchored.cfg","weights/yolo_2k_reanchored_70000.weights","cfg/2k_aug.data") #New yolov3 model #self.yolov3 = YOLOV3("cfg/yolov3_rbc.cfg","weights/yolov3_rbc_60000.weights","cfg/2k_aug.data") self.below_line=set() self.above_line=set() self.considered=set() self.speed1=set() self.idtocoords=dict() self.classes = ['car','bus','two-wheeler','three-wheeler','people','lcv','bicycle','truck'] self.v_count_up= defaultdict(int) self.v_count_down= defaultdict(int) #Initialize each value to zero for c in self.classes: self.v_count_up[c] self.v_count_down[c] self.sline1=sline1 self.sline2=sline2 self.line=line self.tot_dist=0 self.tot_time=0 self.speed=0 self.speed_dict = defaultdict(int) self.frame_count = defaultdict(int)
def freeze_graph(checkpoint_path, output_node_names, savename): with tf.name_scope('input'): input_data = tf.placeholder(dtype=tf.float32, shape=(1, INPUTSIZE, INPUTSIZE, 3), name='input_data') training = tf.placeholder(dtype=tf.bool, name='training') prefixdict = collectpth(checkpoint_path) output = YOLOV3(training).build_network_dynamic(input_data, prefixdict, inputsize=INPUTSIZE) with tf.Session() as sess: output_graph_def = tf.graph_util.convert_variables_to_constants( sess=sess, input_graph_def=sess.graph_def, output_node_names=output_node_names.split(",")) with tf.gfile.GFile(savename, "wb") as f: f.write(output_graph_def.SerializeToString()) print("%d ops in the final graph." % len(output_graph_def.node))
def __init__(self, line, sline1, sline2): #line is for counting #slines for speed estimation self.deepsort = deepsort_rbc() self.yolov3 = YOLOV3( "/home/shantam/alexDarknet/darknet/weights_2903/yolov3_rbc.cfg", "/home/shantam/alexDarknet/darknet/weights_2903/weights2/yolov3_rbc_180000.weights", "/home/shantam/alexDarknet/darknet/weights_2903/ta.data") #New yolov3 model #self.yolov3 = YOLOV3("cfg/yolov3_rbc.cfg","weights/yolov3_rbc_60000.weights","cfg/2k_aug.data") #self.below_line=set() #self.above_line=set() #self.considered=set() #self.speed1=set() #self.idtocoords=dict() self.classes = [ 'car', 'bus', 'two-wheeler', 'three-wheeler', 'people', 'lcv', 'bicycle', 'truck' ]
default='streams.txt', help='source') parser.add_argument( "-fi", "--face_identified", help= "Tipo do identificador de faces utilizado: yolov3, mtcnn,mtcnno,retina", type=str, default='yolov3') args = parser.parse_args() conf = get_config(False) mtcnn = MTCNN() print('mtcnn loaded') yolov3 = YOLOV3() print("Yolov3 loaded") mtcnno = MTCNN_O() print("mtcnno loaded") retinanet = RETINANET("mobile0.25") print("retinanet loaded") learner = face_learner(conf, True) learner.threshold = args.threshold if conf.device.type == 'cpu': learner.load_state(conf, 'cpu_final.pth', True, True) else: learner.load_state(conf, 'mobilefacenet.pth', True, True) learner.model.eval() print('learner loaded')
heatmap_output, offset_output, wh_output = net( torch.autograd.Variable(torch.randn(3, 3, image_h, image_w))) annotations = torch.FloatTensor([[[113, 120, 183, 255, 5], [13, 45, 175, 210, 2]], [[11, 18, 223, 225, 1], [-1, -1, -1, -1, -1]], [[-1, -1, -1, -1, -1], [-1, -1, -1, -1, -1]]]) decode = CenterNetDecoder(image_w, image_h) batch_scores3, batch_classes3, batch_pred_bboxes3 = decode( heatmap_output, offset_output, wh_output) print("3333", batch_scores3.shape, batch_classes3.shape, batch_pred_bboxes3.shape) from yolov3 import YOLOV3 net = YOLOV3(backbone_type="darknet53") image_h, image_w = 416, 416 obj_heads, reg_heads, cls_heads, batch_anchors = net( torch.autograd.Variable(torch.randn(3, 3, image_h, image_w))) annotations = torch.FloatTensor([[[113, 120, 183, 255, 5], [13, 45, 175, 210, 2]], [[11, 18, 223, 225, 1], [-1, -1, -1, -1, -1]], [[-1, -1, -1, -1, -1], [-1, -1, -1, -1, -1]]]) decode = YOLOV3Decoder(image_w, image_h) batch_scores4, batch_classes4, batch_pred_bboxes4 = decode( obj_heads, reg_heads, cls_heads, batch_anchors) print("4444", batch_scores4.shape, batch_classes4.shape, batch_pred_bboxes4.shape)
# def load_model(cfg_path,weights_path,data_path): # net = load_net(cfg_path,weights_path, 0) # meta = load_meta(data_path) if __name__ == '__main__': mp.set_start_method('spawn') # establish queues m = mp.Manager() lock = m.Lock() tasks = mp.JoinableQueue() results = mp.Queue() filepath = sys.argv[1] yv3 = YOLOV3("cfg/yolo_2k_reanchored.cfg", "weights/yolo_2k_reanchored_70000.weights", "cfg/2k_aug.data") # fchkr = folder_checker() #start consumers # number_consumers = mp.cpu_count() * 2 number_consumers = 2 print("cpu count = {}".format(number_consumers)) consumers = [yolo_load(tasks, results) for i in range(number_consumers)] # print(consumers) for w in consumers: # print(w) w.start() streams = glob.glob(os.path.join(filepath, '*')) for strm in streams: