Abstract
Multiple-object tracking (MOT) is a crucial component in autonomous driving systems. However, inaccurate object detection is always the bottleneck for MOT. Most detectors are not designed to take the temporal information across consecutive frames into consideration. To take advantage of such information, we design a novel data representation, the spatio-temporal (ST) map, which collects a batch of detection results spatio-temporally, and we train a novel network, ST-TrackNet, to assign predicted track IDs to each positive detection across a sequence. With our ST map detection fed into the tracker, the correlation of objects between adjacent frames becomes prominent, which improves the performance of the tracker in the data association step. Moreover, the long-term trajectory in a sequence also helps to refine the detection results. We train and evaluate our network on the KITTI dataset, a CARLA simulation dataset, and a dataset recorded in a factory environment. Our approach generally achieves superior performance over the state-of-the-art. © 2022 IEEE.
Original language | English |
---|---|
Pages (from-to) | 284-295 |
Journal | IEEE Transactions on Automation Science and Engineering |
Volume | 21 |
Issue number | 1 |
Online published | 31 Oct 2022 |
DOIs | |
Publication status | Published - Jan 2024 |
Externally published | Yes |
Research Keywords
- Autonomous driving
- deep learning
- Detectors
- Feature extraction
- Long short term memory
- multi-object tracking
- Object detection
- Point cloud compression
- spatio-temporal map
- Tracking
- Trajectory