Multi-Object Tracking and Segmentation on BDD MOTS (test)
49.7mIDF1Ours-SAM-H
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Ours-SAM-HTraining Mode=Self-supervised, Zero-shot=true, Backbone=ViT-H2024.06 | 49.7 | 54.5 | 54.7 | 35.8 | 40.8 | |
| Ours-DeticTraining Mode=Self-supervised, Zero-shot=true, Backbone=Swin-B2024.06 | 49.5 | 53.5 | 54.4 | 36.4 | 40.2 | |
| Ours-SAM-BTraining Mode=Self-supervised, Zero-shot=true, Backbone=ViT-B2024.06 | 49.2 | 53.9 | 54.8 | 35.2 | 40.7 | |
| Ours-R50+Training Mode=Self-supervised, Zero-shot=true, Backbone=ResNet-50, Pre-training=ImageNet classification2024.06 | 48.7 | 53.6 | 54.7 | 35.2 | 40.4 | |
| Ours-Grounding-DINOTraining Mode=Self-supervised, Zero-shot=true, Backbone=Swin-B2024.06 | 48.6 | 52.3 | 54 | 36.1 | 40 | |
| UNINEXT-HTraining Mode=Fully-supervised, In-domain=true2024.06 | 48.5 | 53.2 | 53.6 | 35.7 | 40.6 |