Accompanying code for paper "DAPO: Defect-aware Hybrid Prompt Optimization via Progressive Tuning for Zero-Shot Multi-type Anomaly Detection and Segmentation" which has been accepted by ECCV 2026
Python-based industrial visual anomaly detection using MVTec datasets, pretrained ResNet18 features, PatchCore-style modeling, ONNX export, and reproducible evaluation.
Docker Compose orchestration for the Industrial Visual Anomaly Detection backend and Python inference service.
End-to-end industrial visual anomaly detection using MVTec AD, ResNet patch-memory features, heatmaps, few-shot defect classification, robustness testing, CLI inference, and Streamlit demo.
To associate your repository with the industial-ai topic, visit your repo's landing page and select "manage topics."