Hybrid Structure-from-Motion and Camera Relocalization for Enhanced Egocentric Localization

TL;DR

This paper introduces EgoLoc-v1, a hybrid SfM and camera relocalization pipeline that improves egocentric localization accuracy by combining 3D scans and 2D video frame matching, outperforming previous methods.

Contribution

The paper proposes a novel ensemble strategy and hybrid approach combining SfM and 2D-3D matching for egocentric camera pose estimation, achieving state-of-the-art results.

Findings

Achieves 1.5% higher success rate than previous state-of-the-art.

Utilizes a hybrid SfM and relocalization pipeline for better pose estimation.

Demonstrates significant improvement in overall success rate.

Abstract

We built our pipeline EgoLoc-v1, mainly inspired by EgoLoc. We propose a model ensemble strategy to improve the camera pose estimation part of the VQ3D task, which has been proven to be essential in previous work. The core idea is not only to do SfM for egocentric videos but also to do 2D-3D matching between existing 3D scans and 2D video frames. In this way, we have a hybrid SfM and camera relocalization pipeline, which can provide us with more camera poses, leading to higher QwP and overall success rate. Our method achieves the best performance regarding the most important metric, the overall success rate. We surpass previous state-of-the-art, the competitive EgoLoc, by $1.5\%$. The code is available at \url{https://github.com/Wayne-Mai/egoloc_v1}.