VTAB

Benchmarks

Task Name	Dataset Name	SOTA Result
Image Classification	VTAB 1k	Overall Mean Accuracy79	258
Image Classification	VTAB 1k (test)	Accuracy (Natural)89.09	121
Image Classification	VTAB	Overall Accuracy91.97	103
Image Classification	VTAB-1K 1.0 (test)	Natural Accuracy86	102
Visual Task Adaptation	VTAB 1K	Average Accuracy78.02	78
Image Classification	VTAB v2 (test)	Mean Accuracy70.4	39
Class-Incremental Learning	VTAB B0 Inc10	Last Accuracy94.96	38
Visual Task Adaptation	VTAB	VTAB Mean Accuracy62.66	31
Class-Incremental Learning	VTAB	Avg Accuracy92.6	31
Image Classification	VTAB-6	Accuracy89.8	29
Visual Task Adaptation	VTAB-1k v1 (test)	Mean Accuracy79.1	29
Image Classification	VTAB+ 1.0 (test)	Top-1 Accuracy38.9	24
Image Classification	vtab pets	Accuracy97.339	22
Continual Learning	VTAB	Average Training Time per Task1.7	21
Visual Adaptation	VTAB 1k (test)	Natural Accuracy82.5	16
Visual Task Adaptation	VTAB 1K (test)	CIFAR-100 Accuracy78.8	15
Class-Incremental Learning	VTAB Free-Flow	AT Score69.38	14
Class-Incremental Learning	VTAB CIL, T=5 (test)	Avg Accuracy92.2	11
Image Classification	VTAB-1K few-shot	Caltech101 Accuracy93.34	10
Image Classification	VTAB	CIFAR-100 Accuracy93.19	10
Class-Incremental Learning	VTAB5T small Uniformly Abrupt scenario	Average Accuracy94.24	10
Class-incremental learning	VTAB T=5 (test)	Final Accuracy94.29	10
Class-Incremental Learning	VTAB-Sim50 Varying scenario	Average Accuracy95.89	9
Class-Incremental Learning	VTAB5T-large Uniformly Abrupt scenario	Average Accuracy89.37	9
Continual Learning	VTAB	Memory Usage (GB)4.3	9

Showing 25 of 46 rows