Deep One-Class Classification via Interpolated Gaussian Descriptor

Benchmark Model Rank Results
anomaly-detection-on-fashion-mnistIGD (pre-trained SSL)#3ROC AUC: 94.41
anomaly-detection-on-fashion-mnistIGD (pre-trained ImageNet)#4ROC AUC: 93.57
anomaly-detection-on-fashion-mnistIGD (scratch)#8ROC AUC: 92.01
anomaly-detection-on-hyper-kvasir-datasetIGD#2AUC: 0.939
anomaly-detection-on-lagIGD#2AUC: 0.796
anomaly-detection-on-mnistIGD (pre-trained ImageNet)#1ROC AUC: 99.27
anomaly-detection-on-mnistIGD (scratch)#2ROC AUC: 98.69
anomaly-detection-on-mvtec-adIGD (pre-trained SSL)#83Detection AUROC: 93.4Segmentation AUROC: 93.0
anomaly-detection-on-mvtec-adIGD#84Detection AUROC: 93.4
anomaly-detection-on-mvtec-adIGD (pre-trained ImageNet)#88Detection AUROC: 92.6Segmentation AUROC: 91
anomaly-detection-on-one-class-cifar-10IGD (pre-trained SSL)#14AUROC: 91.25
anomaly-detection-on-one-class-cifar-10IGD (pre-trained ImageNet)#21AUROC: 83.68
anomaly-detection-on-one-class-cifar-10IGD (scratch)#22AUROC: 74.33