TransFusion -- A Transparency-Based Diffusion Model for Anomaly Detection

Benchmark Model Rank Results
depth-anomaly-detection-and-segmentation-on-mvtec-3d-adTransFusion#1Segmentation AUPRO: 0.947Detection AUROC: 0.957
rgb-3d-anomaly-detection-and-segmentation-on-mvtec-3d-adTransFusion#2Detection AUCROC: 0.982Segmentation AUPRO: 0.983
anomaly-detection-on-visaTransFusion#9Detection AUROC: 98.7Segmentation AUPRO (until 30% FPR): 94.7
anomaly-detection-on-mvtec-adTransFusion#32Detection AUROC: 99.4Segmentation AUPRO: 95.3