DiffuBox: Refining 3D Object Detection with Point Diffusion

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Xiangyu, Liu, Zhenzhen, Luo, Katie Z, Datta, Siddhartha, Polavaram, Adhitya, Wang, Yan, You, Yurong, Li, Boyi, Pavone, Marco, Chao, Wei-Lun, Campbell, Mark, Hariharan, Bharath, Weinberger, Kilian Q.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909418782720000
author Chen, Xiangyu
Liu, Zhenzhen
Luo, Katie Z
Datta, Siddhartha
Polavaram, Adhitya
Wang, Yan
You, Yurong
Li, Boyi
Pavone, Marco
Chao, Wei-Lun
Campbell, Mark
Hariharan, Bharath
Weinberger, Kilian Q.
author_facet Chen, Xiangyu
Liu, Zhenzhen
Luo, Katie Z
Datta, Siddhartha
Polavaram, Adhitya
Wang, Yan
You, Yurong
Li, Boyi
Pavone, Marco
Chao, Wei-Lun
Campbell, Mark
Hariharan, Bharath
Weinberger, Kilian Q.
contents Ensuring robust 3D object detection and localization is crucial for many applications in robotics and autonomous driving. Recent models, however, face difficulties in maintaining high performance when applied to domains with differing sensor setups or geographic locations, often resulting in poor localization accuracy due to domain shift. To overcome this challenge, we introduce a novel diffusion-based box refinement approach. This method employs a domain-agnostic diffusion model, conditioned on the LiDAR points surrounding a coarse bounding box, to simultaneously refine the box's location, size, and orientation. We evaluate this approach under various domain adaptation settings, and our results reveal significant improvements across different datasets, object classes and detectors. Our PyTorch implementation is available at \href{https://github.com/cxy1997/DiffuBox}{https://github.com/cxy1997/DiffuBox}.
format Preprint
id arxiv_https___arxiv_org_abs_2405_16034
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DiffuBox: Refining 3D Object Detection with Point Diffusion
Chen, Xiangyu
Liu, Zhenzhen
Luo, Katie Z
Datta, Siddhartha
Polavaram, Adhitya
Wang, Yan
You, Yurong
Li, Boyi
Pavone, Marco
Chao, Wei-Lun
Campbell, Mark
Hariharan, Bharath
Weinberger, Kilian Q.
Computer Vision and Pattern Recognition
Ensuring robust 3D object detection and localization is crucial for many applications in robotics and autonomous driving. Recent models, however, face difficulties in maintaining high performance when applied to domains with differing sensor setups or geographic locations, often resulting in poor localization accuracy due to domain shift. To overcome this challenge, we introduce a novel diffusion-based box refinement approach. This method employs a domain-agnostic diffusion model, conditioned on the LiDAR points surrounding a coarse bounding box, to simultaneously refine the box's location, size, and orientation. We evaluate this approach under various domain adaptation settings, and our results reveal significant improvements across different datasets, object classes and detectors. Our PyTorch implementation is available at \href{https://github.com/cxy1997/DiffuBox}{https://github.com/cxy1997/DiffuBox}.
title DiffuBox: Refining 3D Object Detection with Point Diffusion
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2405.16034