MaskFeat (no extra data, MViT-L)

Reported on 6 benchmarks across 1 task · 1 paper · 4 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision6 results

VideoonKinetics-700
Top-1 Accuracy· 2021-12-16
80.4
best: 85.9 (InternVideo2-6B)
SOTA
Masked Feature Prediction for Self-Supervised Visual Pre-Training arXiv:2112.09133
VideoonKinetics-700
Top-5 Accuracy· 2021-12-16
95.7
best: 96.7 (UMT-L (ViT-L/16))
SOTA
Masked Feature Prediction for Self-Supervised Visual Pre-Training arXiv:2112.09133
VideoonKinetics-600
Top-1 Accuracy· 2021-12-16
88.3
best: 91.9 (InternVideo2-6B)
SOTA
Masked Feature Prediction for Self-Supervised Visual Pre-Training arXiv:2112.09133
VideoonKinetics-600
Top-5 Accuracy· 2021-12-16
98
best: 98.9 (TubeVit-H)
SOTA
Masked Feature Prediction for Self-Supervised Visual Pre-Training arXiv:2112.09133
VideoonKinetics-400
Acc@1· 2021-12-16
86.7
best: 93.6 (OmniVec2)
Masked Feature Prediction for Self-Supervised Visual Pre-Training arXiv:2112.09133
VideoonKinetics-400
Acc@5· 2021-12-16
97.3
best: 98.9 (TubeViT-H (ImageNet-1k))
Masked Feature Prediction for Self-Supervised Visual Pre-Training arXiv:2112.09133