Interactive Image Segmentation on HQSeg-44K (val)
90.755-mIoUSAM-REF
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SAM-REFTrain data=HQSeg-44K, Backbone=ViT-B*, Latency(s)=5.12024.08 | 90.75 | 5.9 | 10.08 | 607 | |
| SAM-REFTrain data=COCO+LVIS, Backbone=ViT-H*, Latency(s)=5.222024.08 | 89.6 | 5.44 | 9.16 | 566 | |
| HQ-SAMTrain data=HQSeg-44K, Backbone=ViT-B*, Latency(s)=4.72024.08 | 89.45 | 6.49 | 10.79 | 671 | |
| FocSAMTrain data=COCO+LVIS, Backbone=ViT-H*, Latency(s)=18.62024.08 | 88.6 | 5.74 | 9.44 | 580 | |
| SAM-REFTrain data=COCO+LVIS, Backbone=ViT-B*, Latency(s)=5.12024.08 | 88.5 | 6.19 | 9.84 | 603 | |
| SAMTrain data=SA-1B, Backbone=ViT-H, Latency(s)=4.212024.08 | 88 | 6.5 | 10.53 | 653 | |
| SAMTrain data=SA-1B, Backbone=ViT-B, Latency(s)=42024.08 | 86.16 | 7.46 | 12.42 | 811 | |
| SegNextTrain data=COCO+LVIS, Backbone=ViT-B, Latency(s)=22.12024.08 | 85.71 | 7.18 | 11.52 | 700 | |
| FocalClickTrain data=COCO+LVIS, Backbone=SegFB3-S2384, Latency(s)=40.142024.08 | 85.45 | 7.03 | 10.74 | 649 | |
| SimpleClickTrain data=COCO+LVIS, Backbone=ViT-B, Latency(s)=88.52024.08 | 85.11 | 7.47 | 12.39 | 797 | |
| FocalClickTrain data=COCO+LVIS, Backbone=SegFB3-S2256, Latency(s)=30.852024.08 | 84.63 | 8.12 | 12.63 | 835 | |
| InterFormerTrain data=COCO+LVIS, Backbone=ViT-B, Latency(s)=30.52024.08 | 82.62 | 7.17 | 10.77 | 658 |