MultiMAE (ViT-B)

Reported on 6 benchmarks across 2 tasks · 1 paper · 2 SOTA

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

Medical3 results

Semantic SegmentationonHypersim
mIoU· 2022-04-04
37
best: 49.74 (EMSANet (2x ResNet-34 NBt1D))
SOTA
MultiMAE: Multi-modal Multi-task Masked Autoencoders arXiv:2204.01678
Semantic SegmentationonADE20K val
mIoU· 2022-04-04
46.2
best: 62.8 (BEiT-3)
MultiMAE: Multi-modal Multi-task Masked Autoencoders arXiv:2204.01678
Semantic SegmentationonADE20K
Validation mIoU· 2022-04-04
46.2
best: 63.6 (ViT-P (InternImage-H))
MultiMAE: Multi-modal Multi-task Masked Autoencoders arXiv:2204.01678

Audio3 results

10-shot image generationonHypersim
mIoU· 2022-04-04
37
best: 49.74 (EMSANet (2x ResNet-34 NBt1D))
SOTA
MultiMAE: Multi-modal Multi-task Masked Autoencoders arXiv:2204.01678
10-shot image generationonADE20K val
mIoU· 2022-04-04
46.2
best: 62.8 (BEiT-3)
MultiMAE: Multi-modal Multi-task Masked Autoencoders arXiv:2204.01678
10-shot image generationonADE20K
Validation mIoU· 2022-04-04
46.2
best: 63.6 (ViT-P (InternImage-H))
MultiMAE: Multi-modal Multi-task Masked Autoencoders arXiv:2204.01678