GitHub - xiaozhen228/VCP-CLIP: (ECCV 2024) VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentation
需要构建正样本,异常样本,以及对应的Mask