AI RESEARCH
PANC: Prior-Aware Normalized Cut via Anchor-Augmented Token Graphs
arXiv CS.CV
•
ArXi:2602.06912v2 Announce Type: replace Unsupervised segmentation from self-supervised ViT patches holds promise but lacks robustness: multi-object scenes confound saliency cues, and low-semantic images weaken patch relevance, both leading to erratic masks. To address this, we present Prior-Aware Normalized Cut (PANC), a