Patch-aware Vector Quantized Codebook Learning for Unsupervised Visual Defect Detection
This addresses industrial quality control needs with an incremental improvement in detection accuracy.
The paper tackles unsupervised visual defect detection by proposing a patch-aware dynamic code assignment scheme within a VQ-VAE framework, achieving state-of-the-art performance on datasets like MVTecAD, BTAD, and MTSD.
Unsupervised visual defect detection is critical in industrial applications, requiring a representation space that captures normal data features while detecting deviations. Achieving a balance between expressiveness and compactness is challenging; an overly expressive space risks inefficiency and mode collapse, impairing detection accuracy. We propose a novel approach using an enhanced VQ-VAE framework optimized for unsupervised defect detection. Our model introduces a patch-aware dynamic code assignment scheme, enabling context-sensitive code allocation to optimize spatial representation. This strategy enhances normal-defect distinction and improves detection accuracy during inference. Experiments on MVTecAD, BTAD, and MTSD datasets show our method achieves state-of-the-art performance.