Implementation of CVPR'23 paper "WinCLIP: Zero-/few-shot anomaly classification and segmentation". It successfully reproduces the same zero-/few-shot AD performance as that in the original paper.

Code for "Zero-shot Generalist Graph Anomaly Detection with Unified Neighborhood Prompts"

Reviews of papers on zero-shot anomaly detection using vision-Language models

Official implementation for paper "Anomalyclip: Object-agnostic prompt learning for zero-shot anomaly detection" (ICLR 2024)