jackeasson / slpnet_pytorch Goto Github PK

您好：
感谢分享！
我有个问题需要请教下：
既然 检测网络的输入是512，为何在default_image_preprocess（）函数 resize到1024而不是直接resize到 512呢？
多谢

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xcd in position 32: invalid continuation byte
How to deal with?

您好，我想问一下，制作数据集的时候是直接根据image的名字生成label的吗，有没有代码处理流程；还有就是示例中的label为什么和名字里对应的坐标数字不一样

Using this line in the evaluation step results in the wrong accuracy value for the validation dataset. Is it a bug or am I missing some part?

你可以上传一下你的论文吗？

直接运行train时，报错：
Traceback (most recent call last):
File "G:/code/SLPNet_pytorch-master/load_data.py", line 194, in
lp_dataset = LPDataSet(cfg.val_img_folder_path, cfg.val_txt_folder_path)
File "G:/code/SLPNet_pytorch-master/load_data.py", line 135, in init
self.gt_labels = self._data_init()
File "G:/code/SLPNet_pytorch-master/load_data.py", line 141, in _data_init
gt_labels.append(self.label_loader(self.img_names[i], self.txt_path))
File "G:/code/SLPNet_pytorch-master/load_data.py", line 108, in read_txt_info
with open(txt_path, "r", encoding='utf-8') as f_txt:FileNotFoundError: [Errno 2] No such file or directory: './data/val/label\01375-90_90-205&545_420&619-395&613_222&611_223&547_396&549-0_0_27_16_32_24_29-185-21.txt'

请问该怎么解决呀，谢谢LZ

您好，谢谢你的工作。我想要自己训练CCPD数据集，但是CCPD文件名中的坐标与您的Lable里的四个坐标似乎并不相同，请问如何把CCPD数据集坐标转换成该模型可用的坐标。

请问在制作数据集的时候，如果是比较正面的图像，不是从侧面拍摄的图片，利用labelImg标注得到车牌的四个点的坐标，只有四个值xmin,ymin,xmax,ymax,这样得到四个点坐标（xmin,ymin）(xmax,ymin) (xmax,ymax) (xmin,yamx),这样与CCPD数据集的四点坐标值不太一样，这样在训练的时候是否有影响呢？

Hi,

I am trying to understand how the label information is present in the .txt file under data/train/label/. For example, in SLPNet_pytorch/data/train/label/01-89_85-302&497_478&557-478&549_308&552_304&507_474&504-0_0_6_25_25_33_25-74-7.txt, the coordinates are 303 510,472 510,474 554,307 556. Whereas, if I use the information from the CCPD I seem to find it difficult to replicate the values. Please provide some guidance on creating the .txt files. Thank you!

请问代码需要哪些修改才能使模型使用多卡训练呢？
我在train.py文件中直接加入了
device = "cuda:0" if args.cuda: model = model.cuda(device) model = torch.nn.DataParallel(model, device_ids=[0,1,2])
但是在模型验证时报错
Traceback (most recent call last):
File "train.py", line 538, in
main(parser.parse_args())
File "train.py", line 512, in main
model = train(args, model, device) # Train decoder
File "train.py", line 317, in train
obj_num_list, scores_tensor, coordinates_tensor = model(images, mode1='det_only', mode2='eval')
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in call
result = self.forward(*input, **kwargs)
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 153, in forward
return self.gather(outputs, self.output_device)
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 165, in gather
return gather(outputs, output_device, dim=self.dim)
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/parallel/scatter_gather.py", line 68, in gather
res = gather_map(outputs)
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/parallel/scatter_gather.py", line 63, in gather_map
return type(out)(map(gather_map, zip(*outputs)))
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/parallel/scatter_gather.py", line 63, in gather_map
return type(out)(map(gather_map, zip(*outputs)))
File "/home/guest/fhc/anaconda3/envs/slpnet/lib/python3.7/site-packages/torch/nn/parallel/scatter_gather.py", line 63, in gather_map
return type(out)(map(gather_map, zip(*outputs)))
TypeError: zip argument #1 must support iteration

不使用多GPU训练的话，感觉这个模型要训练很长时间

----- TRAINING - EPOCH 40 -----
LEARNING RATE:  0.0033786452453014168
	=>The average detection loss of epoch 40 is 0.6146
	=>The average recognition loss of epoch %d is None, didn't train
----- VALIDATING - EPOCH 40 -----
The validation mode is: detection only.
=> Precision:  0.0
=> Recall:  0.0
=> mGauss: 0.000

The reason for this is that I can't directly use my own Dataset for training?
What parameters need to be adjusted?
Thank you for your reply！

Hello,

I was running the eval.py against an image from ccpd_fn image for batch size of 1, in the line

您好，试了一下，识别效果不错，请问检测和识别算法有相关的参考资料吗？谢谢

初学人工智能不太了解，想问一下，SSNet_framework.py中87行的net = SSDetectionNet()中的SSDetectionNet()函数为什么没有实现呢

博主有遇到过这样的情况吗？

当我用自己的数据训练模型时，模型会在训练之后的验证过程中卡住导致程序无法继续运行，请问这是什么原因呢，有无解决办法？
print("The validation mode is: detection and recognition.") Tp = 0 Tn_1 = 0 Tn_2 = 0 pdb.set_trace() # 调试时运行到这下一步就执行不下去了 for step, (images, point_label_list, lpchar_label_list, lpchar_length_list, name_list) in enumerate(loader_val): step += 1 #卡住 start_time = time.time()