High-Performance Few-Shot Segmentation with Foundation Models: An Empirical Study

arxiv link

Datasets

Please follow HSNet to prepare few-shot segmentation datasets.

Training and Testing

Training

python train.py --benchmark pascal/coco --logpath ./your_path --fold 0/1/2/3 --img_size 420

Testing

python test.py --benchmark pascal/coco --load ./your_path/model.pt --fold 0/1/2/3 --nshot 1/5 --img_size 420

To Do

• Code.
• Weights.
• Training and testing commands.
• Requirments.

References

Thanks to HSNet, a great few-shot segmentation codebase!