Object-Preserving Siamese Network for Single-Object Tracking on Point Clouds
- 23 August 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 3007-3017
Abstract
Undoubtedly, the object is the primary factor in 3D single-object tracking (SOT) tasks. However, prior Siamese-based trackers overlook the adverse effects resulting from randomly dropped object points during backbone sampling, hindering the prediction of accurate bounding boxes (BBoxes). Therefore, developing an approach that maximizes the preservation of object points and their object-aware features is of the utmost significance. To address this, we propose an object-preserving Siamese network ( OPSNet ) that can effectively maintain object integrity and boost tracking performance. First, an object highlighting module amplifies the object-aware features and extracts discriminative features from the template and search area. Next, object-preserving sampling selects object candidates, obtains object-preserving search area seeds, and discards background points that have less impact on tracking. Finally, an object localization network accurately locates 3D BBoxes based on the object-preserving search area seeds. Extensive experiments demonstrate that the performance of OPSNet exceeds the state-of-the-art performance, achieving success gains of $\sim$ 9.4% and $\sim$ 2.5% on theKeywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62173143)
This publication has 33 references indexed in Scilit:
- 3D Siamese Transformer Network for Single Object Tracking on Point Clouds2022
- SPAN: siampillars attention network for 3D object tracking in point cloudsInternational Journal of Machine Learning and Cybernetics, 2022
- High speed long-term visual object tracking algorithm for real robot systemsNeurocomputing, 2021
- VoteNet: A Deep Learning Label Fusion Method for Multi-atlas Segmentation2019
- Leveraging Shape Completion for 3D Siamese Tracking2019
- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection2018
- Multiple People Tracking by Lifted Multicut and Person Re-identification2017
- Built-in Foreground/Background Prior for Weakly-Supervised Semantic Segmentation2016
- Online Object Tracking: A Benchmark2013
- Are we ready for autonomous driving? The KITTI vision benchmark suite2012