Scalable Benchmarking and Robust Learning for Noise-Free Ego-Motion and 3D Reconstruction from Noisy Video

Xiaohao Xu, Tianyi Zhang, Shibo Zhao, Xiang Li, Sibo Wang, Yongqi Chen, Ye Li, Bhiksha Raj, Matthew Johnson-Roberson, Sebastian Scherer, Xiaonan Huang · 2025

Read Scalable Benchmarking and Robust Learning for Noise-Free Ego-Motion and 3D Reconstruction from Noisy Video by Xiaohao Xu, Tianyi Zhang, Shibo Zhao, Xiang Li, Sibo Wang, Yongqi Chen, Ye Li, Bhiksha Raj, Matthew Johnson-Roberson, Sebastian Scherer, Xiaonan Huang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-14319.txt.

Introduction

Xiaohao Xu and co-authors tackle robot self-location and 3D scene rebuilding from noisy video, in a paper from 2025. Most models in this area are trained and tested on clean data. Real footage, though, has motion blur, sensor flaws, and timing slips, and models trained clean often fail hard in the field.

The team attacks three gaps. First, they built a pipeline that fakes noisy footage at scale, with varied motion and sensor faults. Second, they turned it into a benchmark that exposes how much noise degrades current methods. Third, they propose a test-time method that refines a clean inner 3D model by aligning it with the noisy input. The aim is systems that keep working outside the lab.

Catalog record

CREATORXiaohao Xu, Tianyi Zhang, Shibo Zhao, Xiang Li, Sibo Wang, Yongqi Chen, Ye Li, Bhiksha Raj, Matthew Johnson-Roberson, Sebastian Scherer, Xiaonan Huang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE198 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-14319

Browse the full library · About the Tower of Babel