top of page
Object Tracking
Paper Code. Website
SCATR: Mitigating New Instance Suppression in LiDAR-based Tracking-by-Attention via Second Chance Assignment and Track Query Dropout
Brian Cheong, Letian Wang, Sandro Papais, Steven L. Waslander
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026.
SCATR is a novel LiDAR-based tracking-by-attention framework that addresses the "new instance suppression" problem, which typically causes high false negative errors in joint detection and tracking. The model introduces two architecture-agnostic training strategies, Second Chance Assignment and Track Query Dropout, to better supervise newborn objects and enhance robustness to missing or switched tracks.
JDT3D: Addressing the Gaps in LiDAR-Based Tracking-by-Attention
Brian Cheong, Jiachen (Jason) Zhou and Steven Waslander
European Conference on Computer Vision (ECCV), 2024.
We propose a novel LiDAR-based joint detection and tracking model that leverages transformer-based decoders to propagate object queries over time, implicitly performing object tracking without an association step at inference.
UncertaintyTrack: Exploiting Detection and Localization Uncertainty in Multi-Object Tracking
John (Chang Won) Lee and Steven Waslander
International Conference on Robotics and Automation (ICRA), 2024.
We introduce a collection of extensions that can be applied to tracking-by-detection trackers to account for localization uncertainty estimates from probabilistic object detectors.
SWTrack: Multiple Hypothesis Sliding Window 3D Multi-Object Tracking
Sandro Papais, Robert Ren, and Steven Waslander
International Conference on Robotics and Automation (ICRA), 2024.
We develop a novel multidimensional graph optimization formulation of multiple hypothesis sliding window tracking for mobile robotics applications.
InterTrack: Interaction Transformer for 3D Multi-Object Tracking
John Willes, Cody Reading, Steven L. Waslander
20th Conference on Robots and Vision (CRV). IEEE, 2023.
Our proposed solution, InterTrack, introduces the Interaction Transformer for 3D MOT to generate discriminative object representations for data association.
aUToTrack: A Lightweight Object Detection and Tracking System for the SAE AutoDrive Challenge
Keenan Burnett, Sepehr Samavi, Steven L. Waslander, Timothy D. Barfoot, Angela P. Schoellig
Conference on Robots and Vision (CRV), 2019
We present a new object tracking dataset (UofTPed50), and propose a lightweight object detection and tracking system (aUToTrack) that achieves SOTA performance on the KITTI Object Tracking benchmark.
bottom of page