Towards Object Re-identification from Point Clouds for 3D MOT
Abstract
This thesis studies the problem of object re-identification (ReID) in a 3D multi-object tracking (MOT) context, by learning to match pairs of objects from cropped (e.g., using their predicted 3D bounding boxes) point cloud observations. We are not concerned with state-of-the-art performance for 3D MOT, however. Instead, we seek to answer the following question: In a realistic tracking by-detection context, how does object ReID from point clouds perform relative to ReID from images? To enable such a study, we propose a lightweight matching head that can be concatenated to any set or sequence processing backbone (e.g., PointNet or ViT), creating a family of comparable object ReID networks for both modalities. Run in Siamese style, our proposed point cloud ReID networks can make thousands of pairwise comparisons in real-time (10 Hz). Our findings demonstrate that their performance increases with higher sensor resolution and approaches that of image ReID when observations are sufficiently dense. Additionally, we investigate our network's ability to enhance 3D multi-object tracking, showing that our point cloud ReID networks can successfully re-identify objects that led a strong motion-based tracker into error. To our knowledge, we are the first to study real-time object re-identification from point clouds in a 3D multi-object tracking context.
Collections
Cite this version of the work
Benjamin Thérien
(2023).
Towards Object Re-identification from Point Clouds for 3D MOT. UWSpace.
http://hdl.handle.net/10012/19305
Other formats
Related items
Showing items related by title, author, creator and subject.
-
Identification and Treatment of Critical Contaminants for Managed Aquifer Recharge (MAR) with Reclaimed Water
Yuan, Jie (University of Waterloo, 2015-01-22)Residential and employment growth places great pressure on well-based potable water supply and stream-based wastewater treatment systems, which are common in many areas of Canada. To ensure that increased potable water ... -
Identification of Oil Sands Naphthenic Acid Structures and Their Associated Toxicity to Pimephales promelas and Oryzias latipes
Bauer, Anthony E (University of Waterloo, 2013-01-25)The oil sands, located in north-eastern Alberta, are one of the largest deposits of oil worldwide. Because the Alberta Environmental Protection and Enhancement Act prohibits the release of oil sands process-affected material ... -
Uncertainty Analysis and the Identification of the Contaminant Transport and Source Parameters for a Computationally Intensive Groundwater Simulation
Yin, Yong (University of Waterloo, 2009-05-15)Transport parameter estimation and contaminant source identification are critical steps in the development of a physically based groundwater contaminant transport model. Due to the irreversibility of the dispersion process, ...