VideoMAE V2-g (64x266x266)

Reported on 4 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.

Computer Vision4 results

VideoonKinetics-400
Acc@1· 2023-03-29
90
best: 93.6 (OmniVec2)
VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking arXiv:2303.16727
VideoonKinetics-400
Acc@5· 2023-03-29
98.4
best: 98.9 (TubeViT-H (ImageNet-1k))
VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking arXiv:2303.16727
VideoonKinetics-600
Top-1 Accuracy· uses extra data· 2023-03-29
89.9
best: 91.9 (InternVideo2-6B)
VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking arXiv:2303.16727
VideoonKinetics-600
Top-5 Accuracy· uses extra data· 2023-03-29
98.5
best: 98.9 (TubeVit-H)
VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking arXiv:2303.16727