def __init__(self, cfg, weight, batchSize=1): self.det_model = Darknet(cfg) # self.det_model.load_state_dict(torch.load('models/yolo/yolov3-spp.weights', map_location="cuda:0")['model']) self.det_model.load_weights(weight) self.det_model.net_info['height'] = input_size self.det_inp_dim = int(self.det_model.net_info['height']) assert self.det_inp_dim % 32 == 0 assert self.det_inp_dim > 32 if device != "cpu": self.det_model.cuda() inf_time = get_inference_time(self.det_model, height=input_size, width=input_size) flops = print_model_param_flops(self.det_model, input_width=input_size, input_height=input_size) params = print_model_param_nums(self.det_model) print("Detection: Inference time {}s, Params {}, FLOPs {}".format( inf_time, params, flops)) if libtorch: example = torch.rand(2, 3, 224, 224) traced_model = torch.jit.trace(self.det_model, example) traced_model.save("det_lib.pt") self.det_model.eval() self.im_dim_list = [] self.batchSize = batchSize self.mul_img = False
def __init__(self, batchSize=1): self.det_model = Darknet("src/yolo/cfg/yolov3-spp.cfg") self.det_model.load_weights('models/yolo/yolov3-spp.weights') self.det_model.net_info['height'] = config.input_size self.det_inp_dim = int(self.det_model.net_info['height']) assert self.det_inp_dim % 32 == 0 assert self.det_inp_dim > 32 self.det_model.cuda() self.det_model.eval() self.stopped = False self.batchSize = batchSize
def __init__(self, batchSize=1): self.det_model = Darknet(config.yolo_cfg) # self.det_model.load_state_dict(torch.load('models/yolo/yolov3-spp.weights', map_location="cuda:0")['model']) self.det_model.load_weights(config.yolo_model) self.det_model.net_info['height'] = config.input_size self.det_inp_dim = int(self.det_model.net_info['height']) assert self.det_inp_dim % 32 == 0 assert self.det_inp_dim > 32 if device != "cpu": self.det_model.cuda() self.det_model.eval() self.stopped = False self.batchSize = batchSize
def __init__(self, cfg, weight, batchSize=1): self.det_model = Darknet(cfg) self.det_model.load_weights(weight) self.det_model.net_info['height'] = opt.input_size self.det_inp_dim = int(self.det_model.net_info['height']) assert self.det_inp_dim % 32 == 0 assert self.det_inp_dim > 32 if device != "cpu": self.det_model.cuda() inf_time = get_inference_time(self.det_model, height=opt.input_size, width=opt.input_size) flops = print_model_param_flops(self.det_model, input_width=opt.input_size, input_height=opt.input_size) params = print_model_param_nums(self.det_model) print("Detection: Inference time {}s, Params {}, FLOPs {}".format(inf_time, params, flops)) self.det_model.eval() self.im_dim_list = [] self.batchSize = batchSize
def __init__(self, cfg, weight, batchSize=1): self.det_model = Darknet(cfg) # self.det_model.load_state_dict(torch.load('models/yolo/yolov3-spp.weights', map_location="cuda:0")['model']) self.det_model.load_weights(weight) self.det_model.net_info['height'] = config.input_size self.det_inp_dim = int(self.det_model.net_info['height']) assert self.det_inp_dim % 32 == 0 assert self.det_inp_dim > 32 if device != "cpu": self.det_model.cuda() inf_time = get_inference_time(self.det_model, height=config.input_size, width=config.input_size) flops = print_model_param_flops(self.det_model, input_width=config.input_size, input_height=config.input_size) params = print_model_param_nums(self.det_model) print("Detection: Inference time {}s, Params {}, FLOPs {}".format( inf_time, params, flops)) self.det_model.eval() self.im_dim_list = [] self.batchSize = batchSize