Segmentation on UniUS-Bench
92.04Muscle Segmentation AccuracyUSFM
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| USFMType=Foundation Model, #Params=101.2M, GFLOPs=53.07, Latency=3.28ms2025.10 | 92.04 | 93.82 | 81.87 | 83.82 | 84.56 | 90.1 | 79.25 | 71.35 | 97.36 | 90.46 | 86.46 | |
| TinyUSFMType=Foundation Model, #Params=6.2M, GFLOPs=3.45, Latency=0.46ms2025.10 | 91.92 | 93.46 | 80.3 | 82.77 | 82.65 | 88.88 | 80 | 69.57 | 97.44 | 90.84 | 85.78 | |
| URFMType=Foundation Model, #Params=101.2M, GFLOPs=53.07, Latency=3.28ms2025.10 | 91.04 | 93.03 | 78.69 | 82.3 | 83.58 | 86.64 | 77.63 | 67.11 | 96.85 | 90.78 | 84.77 | |
| RWKV-UNetType=Conventional Model, #Params=24.6M, GFLOPs=22.08, Latency=3.18ms2025.10 | 90.77 | 92.88 | 72.9 | 82.99 | 83.57 | 87.63 | 78.85 | 71.02 | 96.37 | 90.28 | 84.73 | |
| SeaFormerType=Lightweight Model, #Params=13.9M, GFLOPs=2.53, Latency=0.89ms2025.10 | 90.49 | 91.98 | 73.67 | 80.06 | 81.03 | 86.94 | 73.78 | 66.77 | 95.85 | 90.13 | 83.07 | |
| SegFormerType=Lightweight Model, #Params=13.7M, GFLOPs=5.07, Latency=1.45ms2025.10 | 90.24 | 89.98 | 76 | 80.34 | 79.17 | 86.17 | 73.65 | 68 | 96.13 | 90.17 | 82.99 | |
| VM-UNetType=Conventional Model, #Params=27.4M, GFLOPs=8.15, Latency=3.21ms2025.10 | 89.19 | 90.3 | 74.93 | 82.2 | 82.41 | 86.56 | 79.07 | 68.83 | 95.82 | 90.24 | 83.96 | |
| ViT-TType=Lightweight Model, #Params=6.2M, GFLOPs=3.45, Latency=0.46ms2025.10 | 88.41 | 87.76 | 61.43 | 70.52 | 76.91 | 73.71 | 72.22 | 64.83 | 96.85 | 88.77 | 78.06 |