Instance Segmentation on PennFudanPed (test)
59.3mAPBLO-Inst
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BLO-InstTotal Param(M)=131.63, Trainable Param(M)=38.66, Train Cost (GPU hours)=0.512026.01 | 59.3 | 86.9 | 67.8 | |
| USISTotal Param(M)=698.10, Trainable Param(M)=57.02, Train Cost (GPU hours)=0.682026.01 | 55.1 | 72.6 | 61 | |
| SAM+BTotal Param(M)=111.08, Trainable Param(M)=111.08, Train Cost (GPU hours)=1.252026.01 | 54.3 | 80.5 | 63.7 | |
| SAM+MTotal Param(M)=112.18, Trainable Param(M)=112.18, Train Cost (GPU hours)=2.482026.01 | 54.1 | 78.9 | 61.3 | |
| RS+QueryTotal Param(M)=100.99, Trainable Param(M)=100.99, Train Cost (GPU hours)=1.432026.01 | 53.3 | 76.3 | 62.4 | |
| Mask RTotal Param(M)=41.71, Trainable Param(M)=41.48, Train Cost (GPU hours)=0.382026.01 | 46.4 | 81.8 | 49 | |
| SOLOTotal Param(M)=46.23, Trainable Param(M)=46.01, Train Cost (GPU hours)=0.432026.01 | 42.3 | 75.7 | 39.7 | |
| RS+AnchorTotal Param(M)=117.05, Trainable Param(M)=117.05, Train Cost (GPU hours)=0.432026.01 | 41.1 | 68.4 | 45.9 |