Variational Voxel Pseudo Image Tracking (2023-02-12T00:00:00.000000Z)

TL;DR

A Variational Neural Network-based version of a Voxel Pseudo Image Tracking method for 3D Single Object Tracking and it is shown that both methods improve tracking performance, while penalization of uncertain features provides the best uncertainty quality.

Abstract

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.

Authors

Alexandros Iosifidis

8 papers

Illia Oleksiienko

3 papers

Paraskevi Nousi

2 papers

TL;DR

Abstract

Authors

References41 items

Bayesian learning for neural networks: an algorithmic survey

Variational Neural Networks implementation in Pytorch and JAX

Real-Time 3D Single Object Tracking With Transformer

Variational Neural Networks

Vpit: real-time embedded single object 3D tracking using voxel pseudo images

PTT: Point-Track-Transformer Module for 3D Single Object Tracking in Point Clouds

Box-Aware Feature Enhancement for Single Object Tracking on Point Clouds

Improving convolutional neural networks performance for image classification using test time augmentation: a case study using MURA dataset

Epistemic Neural Networks

A survey of uncertainty in deep neural networks

3D-SiamRPN: An End-to-End Learning Method for Real-Time 3D Single Object Tracking Using Raw Point Cloud

Uncertainty-Aware Voxel based 3D Object Detection and Tracking with von-Mises Loss

Uncertainty-aware Self-supervised 3D Data Association

P2B: Point-to-Box Network for 3D Object Tracking in Point Clouds

Learning an Uncertainty-Aware Object Detector for Autonomous Driving

Deep Sub-Ensembles for Fast Uncertainty Estimation in Image Classification

A General Framework for Uncertainty Estimation in Deep Learning

Uncertainty Estimation in One-Stage Object Detection

Efficient Bird Eye View Proposals for 3D Siamese Tracking

LaserNet: An Efficient Probabilistic 3D Object Detector for Autonomous Driving

SiamRPN++: Evolution of Siamese Visual Tracking With Very Deep Networks

PointPillars: Fast Encoders for Object Detection From Point Clouds

Automatic Brain Tumor Segmentation using Convolutional Neural Networks with Test-Time Augmentation

Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks

Randomized Prior Functions for Deep Reinforcement Learning

Evidential Deep Learning to Quantify Classification Uncertainty

High Performance Visual Tracking with Siamese Region Proposal Network

Towards Safe Autonomous Driving: Capture Uncertainty in the Deep Neural Network For Lidar 3D Vehicle Detection

Fully-Convolutional Siamese Networks for Object Tracking

Learning to Track at 100 FPS with Deep Regression Networks

Object Tracking Benchmark

Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

Weight Uncertainty in Neural Networks

A Novel Performance Evaluation Methodology for Single-Target Trackers

Are we ready for autonomous driving? The KITTI vision benchmark suite

Tracking 3D human pose with large root node uncertainty

Visual object tracking using adaptive correlation filters

UAST: Uncertainty-Aware Siamese Tracking

A New Approach to Linear Filtering and Prediction Problems

A New Approach to Linear Filtering and Prediction Problems

Ieee Transactions on Pattern Analysis and Machine Intelligence High-speed Tracking with Kernelized Correlation Filters

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names