Data Enhancement

[github-luffy]

Using a random cropping method, what is the difference between training and testing？

[isJHan]

Using a random cropping method, what is the difference between training and testing？

You can read dataset.DatasetTrain and dataset.DatasetVal. When training, we crop the input to 256*256 randomly because of the limitation of GPU performance. When testing, we will remain the shape of input because of our desire for accuracy. You can try using DatasetVal when training. When I do this, I found an exception (CUDA out of memory).