|
| Author/organization | Years | Technical characteristics | Dataset | MOTA↑(%) | MOTP↑(%) | IDs↓ |
|
| Ji Zhu/Shanghai Jiao Tong University | 2018 | Unified framework, including SOT and data association [48] | MOT16 | 46.1 | 73.8 | 532 |
| MOT17 | 48.2 | 75.9 | 2194 |
| Peng Chu/Temple University | 2019 | Feature extraction, affinity estimation, and multidimensional distribution are integrated in a single network [49] | MOT2015 | 40.6 | 71.1 | 778 |
| MOT2017 | 52.0 | 76.5 | 3072 |
| KITTI-Car | 77.1 | 77.8 | 123 |
| UA-DETRAC | 19.8 | 36.7 | 617 |
| Weitao Feng/SenseTime Group Limited | 2019 | SOT captures short-term cues, ReID extracts long-term cues [50] | MOT2016 | 49.2 | 74.0 | 606 |
| MOT2017 | 52.7 | 76.2 | 2167 |
| MOT2016p | 69.6 | 78.5 | 768 |
| ShiJie Sun/University of Western Australia | 2019 | End-to-end data association is divided into two stages: feature extraction and affinity estimation [51] | MOT2017 | 52.4 | 52.4 | 8431 |
| MOT2015 | 38.3 | 38.3 | 1648.08 |
| UA-DETRAC | 20.0 | 20.0 | 518.2 |
| Guang Han/Nanjing University of Posts and Telecommunications | 2019 | End-to-end data association, extract features that merge high-level and low-level semantic information [52] | MOT2015 | 38.41 | 72.03 | 1547 |
| MOT2017 | 48.8 | 76.9 | 5601 |
| Brasó/Technical University of Munich | 2019 | End-to-end data association, combining graphs and minimizing costs [53] | 2DMOT2015 | 48.3 | — | 504 |
| MOT2016 | 55.9 | — | 431 |
| MOT2017 | 55.7 | — | 1433 |
| Yihong Xu/Univ. Grenoble Alpes | 2019 | Propose a differentiable tracker that directly optimizes MOTA and MOTP [54] | MOT2017 | 44.3 | 76.0 | 4861 |
| 2019 | End-to-end data association, a deep Hungarian network is proposed for microcomputable MOTA and MOTP [55] | MOT2017 | 53.7 | 77.2 | 1947 |
| MOT2016 | 54.8 | 77.5 | 645 |
| Cong Ma/Peking University | 2019 | End-to-end; combines CNN, ME, and GNN [56] | DukeMTMC | 86.7 | — | 928 |
| MOT16 | 48.6 | — | 594 |
| Yanjie Zeng/South China University of Technology | 2020 | Multivehicle tracking with Wasserstein association metric method [65] | VECHSV | 72.4 | 85.3 | 670 |
| DETRAC | 68.5 | 86.5 | 151 |
|
