Semantic Segmentation on SceneParse150
34.14mIoUFP32
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| FP32Backbone=ViT-B2026.05 | 34.14 | — | — | 12 | 18,900,000,000,000 | |
| MAQEEBackbone=ViT-B2026.05 | 31.2 | 5.74 | 260,000,000,000 | 6.79 | 286,000,000,000 | |
| ViT-SApprox.=original2024.10 | 31 | — | — | — | — | |
| TOASTApprox.=2 → 3, Backbone=ViT-S, Approximator Type=Linear2024.10 | 30 | — | — | — | — | |
| TOASTApprox.=2 → 3, Backbone=ViT-S, Approximator Type=MLP2024.10 | 30 | — | — | — | — | |
| TOASTApprox.=3 → 4, Backbone=ViT-S, Approximator Type=Linear2024.10 | 30 | — | — | — | — | |
| TOASTApprox.=4 → 5, Backbone=ViT-S, Approximator Type=Linear2024.10 | 30 | — | — | — | — | |
| TOASTApprox.=10 → 11, Backbone=ViT-S, Approximator Type=Linear2024.10 | 30 | — | — | — | — | |
| TOASTApprox.=0 → 1, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=0 → 1, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=1 → 2, Backbone=ViT-S, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=1 → 2, Backbone=ViT-S, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=1 → 2, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=1 → 2, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=2 → 3, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=2 → 3, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=3 → 4, Backbone=ViT-S, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=3 → 4, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=3 → 4, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=4 → 5, Backbone=ViT-S, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=4 → 5, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=4 → 5, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=5 → 6, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=5 → 6, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=6 → 7, Backbone=DINO-B, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=6 → 7, Backbone=DINO-B, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=7 → 8, Backbone=ViT-S, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=9 → 10, Backbone=ViT-S, Approximator Type=Linear2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=9 → 10, Backbone=ViT-S, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=10 → 11, Backbone=ViT-S, Approximator Type=MLP2024.10 | 29 | — | — | — | — | |
| DINO-BApprox.=original2024.10 | 29 | — | — | — | — | |
| TOASTApprox.=5 → 6, Backbone=ViT-S, Approximator Type=Linear2024.10 | 28 | — | — | — | — | |
| TOASTApprox.=6 → 7, Backbone=ViT-S, Approximator Type=Linear2024.10 | 28 | — | — | — | — | |
| TOASTApprox.=7 → 8, Backbone=ViT-S, Approximator Type=MLP2024.10 | 28 | — | — | — | — | |
| TOASTApprox.=8 → 9, Backbone=ViT-S, Approximator Type=Linear2024.10 | 28 | — | — | — | — | |
| TOASTApprox.=8 → 9, Backbone=DINO-B, Approximator Type=Linear2024.10 | 28 | — | — | — | — | |
| ViT-EE+ERQBackbone=ViT-B2026.05 | 27.34 | — | — | 8.1 | 317,000,000,000 | |
| TOASTApprox.=0 → 1, Backbone=ViT-S, Approximator Type=Linear2024.10 | 27 | — | — | — | — | |
| TOASTApprox.=5 → 6, Backbone=ViT-S, Approximator Type=MLP2024.10 | 27 | — | — | — | — | |
| TOASTApprox.=6 → 7, Backbone=ViT-S, Approximator Type=MLP2024.10 | 27 | — | — | — | — | |
| TOASTApprox.=8 → 9, Backbone=ViT-S, Approximator Type=MLP2024.10 | 27 | — | — | — | — | |
| TOASTApprox.=8 → 9, Backbone=DINO-B, Approximator Type=MLP2024.10 | 27 | — | — | — | — | |
| TOASTApprox.=9 → 10, Backbone=DINO-B, Approximator Type=Linear2024.10 | 27 | — | — | — | — | |
| TOASTApprox.=10 → 11, Backbone=DINO-B, Approximator Type=Linear2024.10 | 27 | — | — | — | — | |
| ViT-EE+MPQBackbone=ViT-B2026.05 | 26.52 | 11.46 | 394,000,000,000 | 8.08 | 316,000,000,000 | |
| TOASTApprox.=0 → 1, Backbone=ViT-S, Approximator Type=MLP2024.10 | 26 | — | — | — | — | |
| TOASTApprox.=7 → 8, Backbone=DINO-B, Approximator Type=Linear2024.10 | 26 | — | — | — | — | |
| TOASTApprox.=9 → 10, Backbone=DINO-B, Approximator Type=MLP2024.10 | 26 | — | — | — | — | |
| TOASTApprox.=10 → 11, Backbone=DINO-B, Approximator Type=MLP2024.10 | 26 | — | — | — | — | |
| TOASTApprox.=7 → 8, Backbone=DINO-B, Approximator Type=MLP2024.10 | 23 | — | — | — | — |