Implementation of "Nowhere to Disguise: Spot Camouflaged Objects via Saliency Attribute Transfer" in PyTorch.
Requirements
- PyTorch==1.5.1
- TorchVision==0.6.1
- CUDA==11.0
- tqdm==4.62.2
Generate images
-
Download 'datasets.zip' at https://pan.baidu.com/s/1wMsTGSbv7Pa_gdZ5GlFoPA Extraction code:fgh2 . Download 'snapshot.zip' at https://pan.baidu.com/s/1OAA-OoRuZfeQACBHueVdZQ Extraction code:a6vu . Then unzip 'datasets.zip' and 'snapshot.zip' then put them under ./SAT/
-
Run ’Gen_imgs.py‘ , the generated weakly camouflaged images will be save in ./SAT/generated_imgs/
-
We also provide weakly camouflaged images at https://pan.baidu.com/s/1swZhd8NEKJR2r6WYwQfPTQ Extraction code:w3jb .