AI RESEARCH
Do Instance Priors Help Weakly Supervised Semantic Segmentation?
arXiv CS.CV
•
ArXi:2604.11170v1 Announce Type: new Semantic segmentation requires dense pixel-level annotations, which are costly and time-consuming to acquire. To address this, we present SeSAM, a framework that uses a foundational segmentation model, i.e. Segment Anything Model (SAM), with weak labels, including coarse masks, scribbles, and points. SAM, originally designed for instance-based segmentation, cannot be directly used for semantic segmentation tasks.