Multi-Object Tracking on DanceTrack
22FPSTBDQ-Net
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TBDQ-Net#Parameters (M)=8.5, GPUs=2×4090, Precision=FP32, Inference Hardware=one NVIDIA 4090, Backbone=pretrained YOLOX2024.11 | 22 | 22 | |
| MOTRv2#Parameters (M)=41.7, GPUs=2×4090, Precision=FP32, Inference Hardware=one NVIDIA 4090, Protocol=Offline detection results2024.11 | 20.5 | 33 | |
| MOTRv2#Parameters (M)=41.7, Precision=FP32, Inference Hardware=one NVIDIA 4090, Protocol=Practical FPS incorporating YOLOX online inference2024.11 | 16 | — | |
| CO-MOT#Parameters (M)=41.7, GPUs=2×4090, Precision=FP32, Inference Hardware=one NVIDIA 40902024.11 | 14.9 | 125 | |
| MOTR#Parameters (M)=43.9, GPUs=8×V100, Precision=FP32, Inference Hardware=one NVIDIA 40902024.11 | 14.8 | 60 | |
| MOTIP#Parameters (M)=58.9, GPUs=8×4090, Precision=FP32, Inference Hardware=one NVIDIA 40902024.11 | 14.4 | 36 |