Models as Tools: An Agentic Coordination Framework for Unified Multimodal Visual Tracking

Benchmark Model Rank Results
rgb-t-tracking-on-lasherACTrack-EM–Precision: 79.4Success: 63.7
visual-object-tracking-on-got-10kACTrack-B 224–Average Overlap: 83.8Success Rate 0.5: 95.4Success Rate 0.75: 82.0
visual-object-tracking-on-got-10kACTrack-L 384–Average Overlap: 86.8Success Rate 0.5: 95.9Success Rate 0.75: 87.2
visual-object-tracking-on-needforspeedACTrack-B 224–AUC: 0.737
visual-object-tracking-on-tnl2kACTrack-EM–AUC: 77.2precision: 84.7
visual-object-tracking-on-uav123ACTrack-B 224–AUC: 0.743