AdaCLIP

Reported on 6 benchmarks across 1 task · 1 paper

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

Methodology6 results

Anomaly DetectiononMPDD
Detection AUROC· 2024-07-22
82.5
best: 99.6 (GLASS)
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection arXiv:2407.15795
Anomaly DetectiononMPDD
Segmentation AUROC· 2024-07-22
96.1
best: 99.4 (GLASS)
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection arXiv:2407.15795
Anomaly DetectiononMVTec AD
Detection AUROC· 2024-07-22
89.2
best: 99.9 (GLASS)
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection arXiv:2407.15795
Anomaly DetectiononMVTec AD
Segmentation AUROC· 2024-07-22
88.7
best: 99.7 (WeakREST-Block)
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection arXiv:2407.15795
Anomaly DetectiononVisA
Detection AUROC· uses extra data· 2024-07-22
85.8
best: 99.8 (UniNet)
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection arXiv:2407.15795
Anomaly DetectiononVisA
Segmentation AUROC· uses extra data· 2024-07-22
95.5
best: 99.1 (Dinomaly ViT-L (model-unified multi-class))
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection arXiv:2407.15795