One Language-Free Foundation Model Is Enough for Universal V
SAM3
Universal Anomaly Segmentation
Zero and Few-Shot Visual Anomaly Detection