Res$^2$CLIP: Few-Shot Generalist Anomaly Detection with Residual-to-Residual Alignment

Researchers introduced Res^2CLIP, a few-shot anomaly detection method built on CLIP.

It keeps optimization in CLIP’s residual space, using text prompts, visual prompts, and a new residual alignment branch.

The approach is designed to reduce cross-granularity mismatch and domain shift during anomaly detection.

Res^2CLIP aims to generalize better to new categories with only a few labeled examples, without retraining CLIP.