Awesome

Omni-supervised Shadow Detection with Vision Foundation Model

by Zeheng Qian, Wen Wu, Xian-Tao Wu, and Xiao-Diao Chen

News: In 2023.11.26, We release the unsorted code for other researchers. The sorted code and datasets will be released after.

Highlights

🏆 Low GPU requirements. (one 3090 with 24G GPU memory is enough)
🏆 Flexible labeling manner for shadow images

Training

Once you have the data ready, you can start training the model.

python train.py --modelname ShadowSAM --task <your dataset config name>

Testing

python test.py --modelname ShadowSAM --task <your dataset config name>