Bootstrap Fine-Grained Vision-Language Alignment for Unified Zero-Shot Anomaly Localization
This work addresses the challenge of fine-grained anomaly detection for applications like industrial inspection, though it is incremental as it builds on existing CLIP models.
The paper tackles the problem of precise visual anomaly localization in zero-shot settings by proposing AnoCLIP, which introduces a training-free attention mechanism and domain-aware prompting to align patch-level vision with text, achieving significant improvements on various datasets.
Contrastive Language-Image Pre-training (CLIP) models have shown promising performance on zero-shot visual recognition tasks by learning visual representations under natural language supervision. Recent studies attempt the use of CLIP to tackle zero-shot anomaly detection by matching images with normal and abnormal state prompts. However, since CLIP focuses on building correspondence between paired text prompts and global image-level representations, the lack of fine-grained patch-level vision to text alignment limits its capability on precise visual anomaly localization. In this work, we propose AnoCLIP for zero-shot anomaly localization. In the visual encoder, we introduce a training-free value-wise attention mechanism to extract intrinsic local tokens of CLIP for patch-level local description. From the perspective of text supervision, we particularly design a unified domain-aware contrastive state prompting template for fine-grained vision-language matching. On top of the proposed AnoCLIP, we further introduce a test-time adaptation (TTA) mechanism to refine visual anomaly localization results, where we optimize a lightweight adapter in the visual encoder using AnoCLIP's pseudo-labels and noise-corrupted tokens. With both AnoCLIP and TTA, we significantly exploit the potential of CLIP for zero-shot anomaly localization and demonstrate the effectiveness of AnoCLIP on various datasets.