DN2 CLIP ViT

Reported on 2 benchmarks across 1 task · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology2 results

Anomaly DetectiononAnomaly Detection on Anomaly Detection on Unlabeled ImageNet-30 vs Flowers-102
ROC-AUC· uses extra data· 2020-02-24
93.2
best: 98.3 (PsudoLabels CLIP ViT)
SOTA
Deep Nearest Neighbor Anomaly Detection arXiv:2002.10445
Anomaly DetectiononAnomaly Detection on Unlabeled ImageNet-30 vs CUB-200
ROC-AUC· uses extra data· 2020-02-24
93.8
best: 99.4 (PsudoLabels CLIP ViT)
SOTA
Deep Nearest Neighbor Anomaly Detection arXiv:2002.10445