Share your thoughts, 1 month free Claude Pro on usSee more

ImageNet-R

Benchmarks

Task Name	Dataset Name	SOTA Result
Class-Incremental Learning	ImageNet-R B0 Inc20	Last Accuracy83.87	79
Image Classification	ImageNet-R 1.0 (val)	Top-1 Acc71.1	58
Class-Incremental Learning	Split ImageNet-R	Average Forgetting Measure1.27	57
Class-incremental learning	ImageNet-R 10-task	FAA82.06	54
Continual Learning	ImageNet-R 10 tasks	Average ACC@1087.18	28
Continual Learning	ImageNet-R 10-task split	FAA82.06	26
Image Classification	ImageNet-R OfficeHome source (test)	Accuracy0.7718	25
Image Classification	ImageNet-R MiniDomainnet source (test)	Accuracy70.18	25
Class-Incremental Learning	Split ImageNet-R 10 incremental tasks	Forward Accuracy75.06	25
Continual Learning	ImageNet-R (20 tasks)	Average Accuracy (20 Tasks)86.26	22
Class-incremental learning	ImageNet-R 20-split	Accuracy79.4	21
Continual Learning	ImageNet-R (test)	Accuracy87.55	20
Class-Incremental Learning	ImageNet-R N = 10	Accuracy80.64	16
Long-Tail Class-Incremental Learning	ImageNet-R 20 tasks rho=0.01	Accuracy (Last Task)70	14
Long-Tail Class-Incremental Learning	ImageNet-R 20 tasks rho=0.1	Accuracy (Last Task)73.7	14
Image Classification	ImageNet-R strong appearance-shift (test)	Accuracy44.2	14
Continual Learning	ImageNet-R 10 tasks 1.0 (test)	ACC1071.84	14
Continual Learning	ImageNet-R 10 sequential tasks 200 classes	A1:N82.92	14
Class Incremental Learning	ImageNet-R Inc5 (test)	Average Accuracy72.52	13
Federated Class-Incremental Learning	ImageNet-R	FAA (β=0.5)75.23	13
Exemplar-Free Class-Incremental Learning	ImageNet-R 5-task (test)	Average Accuracy (AA)49.34	12
Class-Incremental Learning	ImageNet-R 50-task partition	FAA82.06	10
Text-guided Image-to-Image Translation	ImageNet-R TI2I modified	CLIP Similarity32	10
Continual Learning	ImageNet-R (20-Split)	Accuracy74.73	9
Continual Learning	ImageNet-R (5-Split)	Accuracy79.88	9

Showing 25 of 53 rows