5 datasets found

The CANet dataset is a benchmark for few-shot semantic segmentation.

The PANet dataset is a benchmark for few-shot semantic segmentation.

The proposed Semantic Meta-Learning (SML) framework for few-shot semantic segmentation.

Few-shot segmentation remains challenging due to the limitations of its labeling information for unseen classes. Most previous approaches rely on extracting high-level feature...

Few-shot semantic segmentation via prototype augmentation with image-level annotations