search

Ugness / PiCANet-Implementation

Pytorch Implementation of PiCANet: Learning Pixel-wise Contextual Attention for Saliency Detection
MIT License
178 stars 40 forks source link

Testing problems #4

Closed githubBingoChen closed 6 years ago

githubBingoChen commented 6 years ago

When I use the testing code, I found that you resize the input image and gt in your class DUTS_dataset in Dataset.py as following self.transform = transforms.Compose( [RandomFlip(0.5), RandomCrop(224), ToTensor()]) Could you check whether this is an error in your testing code?

Ugness commented 6 years ago

I thought that same condition on test and training set is fair, but I think that could be an error if you think that resizing is correct for testing. I'll add an testing transform option. Thanks.