Semantic Segmentation on ADE20K 52 (val)
52.82mIoUFull Dataset
Evaluation Results
| Method | Links | |
|---|---|---|
| Full DatasetArchitecture=Mask2Former, Backbone=Swin-B2026.05 | 52.82 | |
| Full DatasetArchitecture=Mask2Former, Backbone=Swin-S2026.05 | 51.22 | |
| Full DatasetArchitecture=Mask2Former, Backbone=Swin-T2026.05 | 48.92 | |
| Full DatasetArchitecture=SegFormer, Backbone=MiT-B42026.05 | 48.2 | |
| Full DatasetArchitecture=Mask2Former, Backbone=Res-502026.05 | 47.03 | |
| Full DatasetArchitecture=SegFormer, Backbone=MiT-B22026.05 | 45.14 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 32.63 | |
| D3S2Architecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 31.46 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 30.28 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 29.1 | |
| UniformArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 28.46 | |
| D3S2Architecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 28.17 | |
| D3S2Architecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 28.15 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 27.59 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 27.48 | |
| HerdingArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 26.48 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 26.13 | |
| UniformArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 25.22 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 24.99 | |
| D3S2Architecture=Mask2Former, Backbone=Res-50, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 24.67 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 24.61 | |
| RandomArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 24.53 | |
| D3S2Architecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 24.45 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 24 | |
| D3S2Architecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 23.82 | |
| UniformArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 23.48 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 23.19 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 22.81 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 22.7 | |
| K-CenterArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 22.56 | |
| HerdingArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 22.56 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 22.55 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 21.7 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 21.69 | |
| UniformArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 21.63 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 21.42 | |
| D3S2Architecture=Mask2Former, Backbone=Res-50, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 20.96 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 20.74 | |
| RandomArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 20.72 | |
| UniformArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 20.36 | |
| D3S2Architecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 20.28 | |
| HerdingArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 19.88 | |
| RandomArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 19.74 | |
| UniformArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 19.53 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 19.45 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 19.16 | |
| K-CenterArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=2%, Guidance Model=MiT-B22026.05 | 19.14 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 18.98 | |
| K-CenterArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 18.95 | |
| HerdingArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 18.34 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 18.09 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 17.94 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 17.94 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 17.67 | |
| RandomArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 17.5 | |
| UniformArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 17.2 | |
| RandomArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 16.86 | |
| HerdingArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 16.78 | |
| D3S2Architecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 16.59 | |
| UniformArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 16.32 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 16.32 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 16.15 | |
| D3S2Architecture=Mask2Former, Backbone=Res-50, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 15.87 | |
| HerdingArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 15.78 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 15.65 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 15.09 | |
| RandomArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 15.04 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 15.04 | |
| K-CenterArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=1%, Guidance Model=MiT-B22026.05 | 14.9 | |
| K-CenterArchitecture=SegFormer, Backbone=MiT-B4, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 14.83 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 14.67 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 14.43 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 13.76 | |
| K-CenterArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=2%, Guidance Model=Swin-T2026.05 | 13.75 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 13.63 | |
| HerdingArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 13.46 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 13.32 | |
| HerdingArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 13.3 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 13.27 | |
| RandomArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 12.99 | |
| RandomArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 12.59 | |
| UniformArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 12.54 | |
| UniformArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 12.51 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 12.47 | |
| K-CenterArchitecture=SegFormer, Backbone=MiT-B2, Distillation Ratio=0.5%, Guidance Model=MiT-B22026.05 | 12.31 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-B, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 11.92 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 11.14 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 10.98 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-S, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 10.52 | |
| HerdingArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 10.5 | |
| K-CenterArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=1%, Guidance Model=Swin-T2026.05 | 9.88 | |
| HerdingArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 9.75 | |
| RandomArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 9.57 | |
| RandomArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 9.53 | |
| K-CenterArchitecture=Mask2Former, Backbone=Swin-T, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 8.66 | |
| K-CenterArchitecture=Mask2Former, Backbone=Res-50, Distillation Ratio=0.5%, Guidance Model=Swin-T2026.05 | 7.92 |