Active visual tracking on EVT-Bench Single Target single-view
90TRQwen-RobotNav-4B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen-RobotNav-4BModel size=4B2026.06 | 90 | 6.4 | 77.4 | |
| Qwen-RobotNav-8BModel size=8B2026.06 | 89.7 | 5.7 | 78.6 | |
| ABot-N02026.06 | 87.6 | 8.54 | 86.9 | |
| TrackVLA++2026.06 | 81 | 2.1 | 86 | |
| NavFoM2026.06 | 80.5 | — | 85 | |
| TrackVLA2026.06 | 78.6 | 1.65 | 85.1 | |
| IBVSuses GroundingDINO=true2026.06 | 56.2 | 3.75 | 42.9 | |
| EVTuses SoM+GPT-4o=true2026.06 | 49.9 | 40.5 | 32.5 | |
| Uni-NaVid2026.06 | 39.5 | 41.9 | 25.7 | |
| EVT2026.06 | 39.1 | 42.5 | 24.4 | |
| PoliFormeruses GroundingDINO=true2026.06 | 15.5 | 40.1 | 4.67 |