Mask2Former (Swin-L, single scale)

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

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

Computer Vision6 results

Instance SegmentationonCOCO test-dev
APL· 2021-12-02
71.2
best: 72.4 (EVA)
SOTA
Masked-attention Mask Transformer for Universal Image Segmentation arXiv:2112.01527
Instance SegmentationonCOCO test-dev
AP50· 2021-12-02
74.9
best: 80.8 (InternImage-H)
Masked-attention Mask Transformer for Universal Image Segmentation arXiv:2112.01527
Instance SegmentationonCOCO test-dev
AP75· 2021-12-02
54.9
best: 63.4 (Co-DETR)
Masked-attention Mask Transformer for Universal Image Segmentation arXiv:2112.01527
Instance SegmentationonCOCO test-dev
APM· 2021-12-02
53.8
best: 60.1 (Co-DETR)
Masked-attention Mask Transformer for Universal Image Segmentation arXiv:2112.01527
Instance SegmentationonCOCO test-dev
APS· 2021-12-02
29.1
best: 41.6 (Co-DETR)
Masked-attention Mask Transformer for Universal Image Segmentation arXiv:2112.01527
Instance SegmentationonCOCO test-dev
mask AP· 2021-12-02
50.5
best: 57.1 (Co-DETR)
Masked-attention Mask Transformer for Universal Image Segmentation arXiv:2112.01527