PyTorch 1.10.1, torchvision 0.11.2. The code is tested with python=3.8, cuda=10.2.
Download the training sets and unzip them to ./trainsets/Flickr/Train
.
Download the val sets and unzip them to ./testsets/Flickr
.
Generate binaries
We pre-process the images before training. This step will decode all png files and save them as binaries.
python prepare/create_bin.py
Generate metric list by informative importance
python prepare/create_bin_ii.py
Run train.py
to perform training.
Download the test sets and unzip them to ./testsets
.
Run test.py
to perform a demo inference. Results (.png files) will be saved to ./results
.