Awesome

IdeNet

This is the official implement code of paper IdeNet: Identifying camouflaged objects like creatures.

Note

The results and trained model will be available if the paper is accepted.

Train

1. Change --train_path and --test-path in Train.py
   run python Train.py
   The trained model will be save in ./checkpoints/\

2.change the --pth_path as your save model path, and change the value of data_path and save_path.
run python MyTesting.py to get test results.\

3.Navigate to ./evaltools, change the value of --pred_root (the path of save_path in step 2), --GT_root (the path of test datasets), and model_name (the model name your saved, like "IdeNet" in our paper) in eval.py.
run eval.py to get metrics.