Interactive Segmentation on DAVIS From Scratch 585
4.06NoC @ 85%FocalClick-segformerB3-S2
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| FocalClick-segformerB3-S2Backbone=SegFormer-B3, Input Resolution=S2 (256x256), Training Data=COCO+LVIS2022.04 | 4.06 | 5.89 | 11.12 | 43 | 74 | 218 | |
| RITM-hrnet32Backbone=HRNet-32, Training Data=COCO+LVIS2022.04 | 4.74 | 6.74 | 12.09 | 45 | 80 | 230 | |
| FocalClick-hrnet32-S2Backbone=HRNet-32, Input Resolution=S2 (256x256), Training Data=COCO+LVIS2022.04 | 4.77 | 6.84 | 11.9 | 48 | 76 | 241 | |
| FocalClick-hrnet18s-S2Backbone=HRNet-18s, Input Resolution=S2 (256x256), Training Data=COCO+LVIS2022.04 | 4.93 | 6.87 | 11.97 | 49 | 77 | 239 | |
| FocalClick-segformerB0-S2Backbone=SegFormer-B0, Input Resolution=S2 (256x256), Training Data=COCO+LVIS2022.04 | 4.99 | 7.13 | 12.65 | 50 | 86 | 260 | |
| RITM-hrnet18sBackbone=HRNet-18s, Training Data=COCO+LVIS2022.04 | 5.34 | 7.57 | 12.94 | 52 | 91 | 257 | |
| FocalClick-hrnet18s-S1Backbone=HRNet-18s, Input Resolution=S1 (128x128), Training Data=COCO+LVIS2022.04 | 5.62 | 8.08 | 13.73 | 53 | 98 | 274 | |
| FocalClick-segformerB0-S1Backbone=SegFormer-B0, Input Resolution=S1 (128x128), Training Data=COCO+LVIS2022.04 | 6.21 | 9.06 | 14.81 | 64 | 127 | 315 |