AdaMAE

Reported on 4 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.

Computer Vision4 results

VideoonSomething-Something V2
Acc@1· 2022-11-16
70.04
SOTA
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders arXiv:2211.09120
VideoonSomething-Something V2
Acc@5· 2022-11-16
92.7
SOTA
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders arXiv:2211.09120
VideoonKinetics-400
Acc@1· 2022-11-16
81.7
best: 93.6 (OmniVec2)
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders arXiv:2211.09120
VideoonKinetics-400
Acc@5· 2022-11-16
95.2
best: 98.9 (TubeViT-H (ImageNet-1k))
AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning with Masked Autoencoders arXiv:2211.09120