Image Classification on ImageNet-1K 1.0 (val) (Top-1 Accuracy and Latency)
84.5Top-1 AccPRISMamba
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PRISMambaParams=22 M, GFlops=3.9 G, TP(img/s)=30542026.02 | 84.5 | — | |
| PRISMamba (w/o PCF)Params=27 M, GFlops=4.6 G, TP(img/s)=21772026.02 | 84.1 | — | |
| SiMBAParams=27 M, GFlops=5.0 G2026.02 | 84 | — | |
| GroupMambaParams=34 M, GFlops=7.0 G, TP(img/s)=8032026.02 | 83.9 | — | |
| VSSDParams=24 M, GFlops=4.5 G2026.02 | 83.7 | — | |
| SparX-MambaParams=27 M, GFlops=5.2 G, TP(img/s)=13702026.02 | 83.5 | — | |
| DefMambaParams=26 M, GFlops=4.8 G2026.02 | 83.5 | — | |
| MaIRParams=26 M, GFlops=5.4 G2026.02 | 83.1 | — | |
| FractalMambaParams=31 M, GFlops=4.8 G2026.02 | 83 | — | |
| LocalMambaParams=26 M, GFlops=5.7 G2026.02 | 82.7 | — | |
| VMambaParams=30 M, GFlops=5.6 G, TP(img/s)=16862026.02 | 82.6 | — | |
| ZigmaParams=31 M, GFlops=5.1 G2026.02 | 82.4 | — | |
| EfficientVMambaParams=33 M, GFlops=4.0 G2026.02 | 81.8 | — | |
| PlainMambaParams=25 M, GFlops=8.1 G2026.02 | 81.6 | — | |
| QuadMambaParams=31 M, GFlops=5.5 G, TP(img/s)=12522026.02 | 81.4 | — | |
| VimParams=26 M, GFlops=5.3 G, TP(img/s)=8112026.02 | 80.5 | — | |
| FastViT-T12Distillation=true2024.01 | 80.3 | 1.4 | |
| SHViT-S4Distillation=true2024.01 | 80.2 | 1.6 | |
| MobileOne-S4Distillation=false2024.01 | 79.4 | 1.7 | |
| SHViT-S4Distillation=false2024.01 | 79.4 | 1.6 | |
| EfficientFormer-L1Distillation=true2024.01 | 79.2 | 1.5 | |
| EfficientNet-B1Distillation=false2024.01 | 79.1 | 1.8 | |
| FastViT-T12Distillation=false2024.01 | 79.1 | 1.4 | |
| EfficientFormerV2-S1Distillation=true2024.01 | 79 | 1.3 | |
| AdventurerParams=12 M, GFlops=4.2 G, TP(img/s)=27572026.02 | 78.2 | — | |
| MobileViTv2 x 1.0Distillation=false2024.01 | 78.1 | 3.8 | |
| EfficientFormerV2-S1Distillation=false2024.01 | 77.9 | 1.3 | |
| EfficientFormer-L1Distillation=false2024.01 | 77.3 | 1.5 |