Entity Segmentation on EntitySeg high-res (test)
41APeCropFormer
Evaluation results
| Method | Links | |
|---|---|---|
| CropFormerInput Preprocessing=resize (800, 1333) + high-res crop resize (800, 1333), GPU Type=A100-40G, GPU Memory (Train)=11.4G, GPU Memory (Inference)=6.9G, Train Time (GPU day)=9.8, Inference Time (ms)=15032024.02 | 41 | |
| Mask2FormerInput Preprocessing=resize (2700, 4500), GPU Type=A100-80G, GPU Memory (Train)=58.8G, GPU Memory (Inference)=10.4G, Train Time (GPU day)=60.8, Inference Time (ms)=33122024.02 | 40.9 | |
| VITAInput Preprocessing=resize (800, 1333) + high-res crop resize (800, 1333), GPU Type=A100-40G, GPU Memory (Train)=11.9G, GPU Memory (Inference)=6.8G, Train Time (GPU day)=10.2, Inference Time (ms)=15122024.02 | 39.9 | |
| V-Mask2FormerInput Preprocessing=resize (800, 1333) + high-res crop resize (800, 1333), GPU Type=A100-40G, GPU Memory (Train)=10.4G, GPU Memory (Inference)=6.7G, Train Time (GPU day)=6.8, Inference Time (ms)=15032024.02 | 39.7 | |
| Mask2FormerInput Preprocessing=resize (800, 1333), GPU Type=A100-40G, GPU Memory (Train)=5.8G, GPU Memory (Inference)=3.0G, Train Time (GPU day)=2.4, Inference Time (ms)=6372024.02 | 39.5 |