Tianrun Chen, Lanyun Zhu, Chaotao Ding, Runlong Cao, Yan Wang, Zejian Li, Lingyun Sun, Papa Mao, Ying Zang

KOKONI, Moxin Technology (Huzhou) Co., LTD , Zhejiang University, Singapore University of Technology and Design, Huzhou University, Beihang University.

Update on 28 April: We tested the performance of polyp segmentation to show our approach can also work on medical datasets. Update on 22 April: We report our SOTA result based on ViT-H version of SAM (use demo.yaml). We have also uploaded the yaml config for ViT-L and ViT-B version of SAM, suitable GPU with smaller memory (e.g. NVIDIA Tesla V-100), although they may compromise on accuracy.

This code was implemented with Python 3.8 and PyTorch 1.13.0. You can install all the requirements via:

pip install -r requirements.txt

1. Download the dataset and put it in ./load.
2. Download the pre-trained SAM(Segment Anything) and put it in ./pretrained.
3. Training:

CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nnodes 1 --nproc_per_node 4 loadddptrain.py --config configs/demo.yaml

!Please note that the SAM model consume much memory. We use 4 x A100 graphics card for training. If you encounter the memory issue, please try to use graphics cards with larger memory!

1. Evaluation:

python test.py --config [CONFIG_PATH] --model [MODEL_PATH]

CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch train.py --nnodes 1 --nproc_per_node 4 --config [CONFIG_PATH]

python test.py --config [CONFIG_PATH] --model [MODEL_PATH]

To be uploaded

• COD10K
• CAMO
• CHAMELEON

• ISTD

• Kvasir

If you find our work useful in your research, please consider citing:

@misc{chen2023sam,
      title={SAM Fails to Segment Anything? -- SAM-Adapter: Adapting SAM in Underperformed Scenes: Camouflage, Shadow, and More}, 
      author={Tianrun Chen and Lanyun Zhu and Chaotao Ding and Runlong Cao and Shangzhan Zhang and Yan Wang and Zejian Li and Lingyun Sun and Papa Mao and Ying Zang},
      year={2023},
      eprint={2304.09148},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

The part of the code is derived from Explicit Visual Prompt by Weihuang Liu, Xi Shen, Chi-Man Pun, and Xiaodong Cun by University of Macau and Tencent AI Lab.