ViT-P (OneFormer, DiNAT-L)

Reported on 6 benchmarks across 2 tasks · 1 paper

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

Medical3 results

Semantic SegmentationonCOCO (Common Objects in Context)
mIoU· 2025-05-26
68.8
best: 77.2 (HyperSeg)
The Missing Point in Vision Transformers for Universal Image Segmentation arXiv:2505.19795
Semantic SegmentationonADE20K
Params (M)· 2025-05-26
309
best: 3000 (FD-SwinV2-G)
The Missing Point in Vision Transformers for Universal Image Segmentation arXiv:2505.19795
Semantic SegmentationonADE20K
Validation mIoU· 2025-05-26
59.9
best: 63.6 (ViT-P (InternImage-H))
The Missing Point in Vision Transformers for Universal Image Segmentation arXiv:2505.19795

Audio3 results

10-shot image generationonCOCO (Common Objects in Context)
mIoU· 2025-05-26
68.8
best: 77.2 (HyperSeg)
The Missing Point in Vision Transformers for Universal Image Segmentation arXiv:2505.19795
10-shot image generationonADE20K
Params (M)· 2025-05-26
309
best: 3000 (FD-SwinV2-G)
The Missing Point in Vision Transformers for Universal Image Segmentation arXiv:2505.19795
10-shot image generationonADE20K
Validation mIoU· 2025-05-26
59.9
best: 63.6 (ViT-P (InternImage-H))
The Missing Point in Vision Transformers for Universal Image Segmentation arXiv:2505.19795