Command Palette
Search for a command to run...
Jialian Wu Jiale Cao Liangchen Song Yu Wang Ming Yang Junsong Yuan
Abstract
Most online multi-object trackers perform object detection stand-alone in a neural net without any input from tracking. In this paper, we present a new online joint detection and tracking model, TraDeS (TRAck to DEtect and Segment), exploiting tracking clues to assist detection end-to-end. TraDeS infers object tracking offset by a cost volume, which is used to propagate previous object features for improving current object detection and segmentation. Effectiveness and superiority of TraDeS are shown on 4 datasets, including MOT (2D tracking), nuScenes (3D tracking), MOTS and Youtube-VIS (instance segmentation tracking). Project page: https://jialianwu.com/projects/TraDeS.html.
Code Repositories
Benchmarks
| Benchmark | Methodology | Metrics |
|---|---|---|
| instance-segmentation-on-cityscapes | - | - |
| instance-segmentation-on-nuscenes | TraDeS | MOTA: 68.2 |
| multi-object-tracking-on-dancetrack | TraDes | AssA: 25.4 DetA: 74.5 HOTA: 43.3 IDF1: 41.2 MOTA: 86.2 |
| multi-object-tracking-on-mot15 | Baseline+MFW | MOTA: 66.5 |
| multi-object-tracking-on-mot16 | TraDeS | IDF1: 64.7 MOTA: 70.1 |
| multi-object-tracking-on-mot17 | TraDeS | IDF1: 63.9 MOTA: 69.1 |
| multi-object-tracking-on-mots20 | TraDes | IDF1: 58.7 sMOTSA: 50.8 |
| online-multi-object-tracking-on-mot16 | TraDeS | MOTA: 67.7 |
| video-instance-segmentation-on-youtube-vis-1 | TraDeS | AP50: 52.6 AP75: 32.8 mask AP: 32.6 |
Build AI with AI
From idea to launch — accelerate your AI development with free AI co-coding, out-of-the-box environment and best price of GPUs.