Variational Voxel Pseudo Image Tracking
Not on the map. Why?
Abstract (arXiv)
Uncertainty estimation is an important task for critical problems, such as robotics and autonomous driving, because it allows creating statistically better perception models and signaling the model's certainty in its predictions to the decision method or a human supervisor. In this paper, we propose a Variational Neural Network-based version of a Voxel Pseudo Image Tracking (VPIT) method for 3D Single Object Tracking. The Variational Feature Generation Network of the proposed Variational VPIT computes features for target and search regions and the corresponding uncertainties, which are later combined using an uncertainty-aware cross-correlation module in one of two ways: by computing similarity between the corresponding uncertainties and adding it to the regular cross-correlation values, or by penalizing the uncertain feature channels to increase influence of the certain features. In experiments, we show that both methods improve tracking performance, while penalization of uncertain features provides the best uncertainty quality.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 30 of its 41 references that are in the map. First posted on arXiv 12 Feb 2023.
Builds on
30 of 41 references.
References · 30
- 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
- 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,707
- 2018Evidential Deep Learning to Quantify Classification Uncertaintycited by 728
- 2021A Survey of Uncertainty in Deep Neural Networkscited by 675
- 2016Fully-Convolutional Siamese Networks for Object Trackingcited by 566
- 2015Weight Uncertainty in Neural Networkscited by 554
- 2018SiamRPN++: Evolution of Siamese Visual Tracking with Very Deep Networkscited by 445
- 2018Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networkscited by 278
- 2018Randomized Prior Functions for Deep Reinforcement Learningcited by 218
- 2019LaserNet: An Efficient Probabilistic 3D Object Detector for Autonomous Drivingcited by 133
- 2021Epistemic Neural Networkscited by 120
- 2019A General Framework for Uncertainty Estimation in Deep Learningcited by 119
18 more
- 2015A Novel Performance Evaluation Methodology for Single-Target Trackerscited by 113
- 2016Learning to Track at 100 FPS with Deep Regression Networkscited by 89
- 2020P2B: Point-to-Box Network for 3D Object Tracking in Point Cloudscited by 84
- 2018Towards Safe Autonomous Driving: Capture Uncertainty in the Deep Neural Network For Lidar 3D Vehicle Detectioncited by 82
- 2021Box-Aware Feature Enhancement for Single Object Tracking on Point Cloudscited by 66
- 2021PTT: Point-Track-Transformer Module for 3D Single Object Tracking in Point Cloudscited by 59
- 20213D-SiamRPN: An End-to-End Learning Method for Real-Time 3D Single Object Tracking Using Raw Point Cloudcited by 45
- 2018Automatic Brain Tumor Segmentation using Convolutional Neural Networks with Test-Time Augmentationcited by 36
- 2019Uncertainty Estimation in One-Stage Object Detectioncited by 36
- 2019Deep Sub-Ensembles for Fast Uncertainty Estimation in Image Classificationcited by 36
- 2019Learning an Uncertainty-Aware Object Detector for Autonomous Drivingcited by 34
- 2022Real-time 3D Single Object Tracking with Transformercited by 27
- 2022Bayesian Learning for Neural Networks: an algorithmic surveycited by 22
- 2019Efficient Bird Eye View Proposals for 3D Siamese Trackingcited by 18
- 2020Uncertainty-Aware Voxel based 3D Object Detection and Tracking with von-Mises Losscited by 10
- 2022Variational Neural Networkscited by 7
- 2020Uncertainty-aware Self-supervised 3D Data Associationcited by 6
- 2022VPIT: Real-time Embedded Single Object 3D Tracking Using Voxel Pseudo Imagescited by 3
Built on by
Not cited yet.