No such file or directory: 'trained_models/checkpoint.pk'

[yingzhige00]

try change option.python: self.load_model = True to self.load_model = Flase I remember my file is autoly built.

[future013]

Thank you. andI'd like to ask about Traceback (most recent call last): File "E:/pytorchproject/Attention-Based-CycleDehaze/dataset.py", line 150, in loop = tqdm(ITS_train_loader, leave=True) NameError: name 'ITS_train_loader' is not defined

[future013]

and How to train my own dataset

[yingzhige00]

i can't deal with the first problem. my project run well. if you want to train your own datasets,you need to input the same size of images to data/clear and data/hazy.

[yingzhige00]

self.load_model = Flase mean you are training a new checkpoint.pk self.load_model = True mean continue to train it.

[future013]

self.load_model = Flase 表示您正在训练新 checkpoint.pk self.load_model = True 表示继续训练它。

Thank you, and how is the training process visualised? Do I need to install visdom?

[yingzhige00]

I don't know. I install visdom, but the code doesn't ask for it. My results are saved in some folders, like Attention-Based-CycleDehaze/saved_images and Attention-Based-CycleDehaze/test_images. My results did not meet the author's description, But the dehazing effect is ok.

[future013]

FileNotFoundError: [Errno 2] No such file or directory: 'data\clear\18.png.png' Why is the data named in the same way as the ITS dataset still reporting errors?

[yingzhige00]

Emmmm I don't know. But I also meet this. my error is because of unnecessary folede ‘.ipynb_checkpoints’. when I delete it, the code can works fine.

[future013]

Thank you. May I ask how you trained your own dataset?