.png){kind=icon}
.png){kind=icon}
Return
Offset3Net: Simple joint 3D detection and tracking with three-step offset learning
.png){kind=icon}
PRE
AI
DOI：10.1109/TII.2023.3290184.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
LIDAR-based multi-object detection and tracking play fundamental roles in autonomous driving systems. Most existing detection and tracking methods inevitably require complex pairing permutations for object association across frames, making the framework slow. Moreover, the occlusion and viewpoint changes lead to missed and false detection. To solve the above issues, this paper proposes a simple joint 3D detection and tracking approach with three-step offset learning (Offset(3)Net). Specifically, Offset(3)Net incorporates three task-specific output subnetworks to learn three offsets: center offset, motion offset, and association offset. The learning of above offsets eliminates the complex bipartite matching processing. Specifically, the center offset guides the model to generate precise detections, while the motion offset transforms the track from the previous frame to the current frame, and the association offset minimizes the distance between detection and motion-updated track of the same object. Then, a simple read-off operation is conducted for data association on a hybrid-time centerness map, which represents the detections and offset-updated tracks. Additionally, we design an detection feature-enhanced module that captures the temporal coherence of the object motion and appearance information, avoiding the missed and false detection. Experiments on nuScenes have demonstrated the effectiveness of our Offset(3)Net in terms of accuracy and speed compared with most 3D detection and tracking methods.
Keywords:
3D LIDAR tacking
3D joint detection and tracking
3D object detection
offset learning
data association
Journal
IF:
9.9
Papers:
8.3K
Citations:
6.0W