Relative Pose Regression with Pose Auto-Encoders: Enhancing Accuracy and Data Efficiency for Retail Applications

2025-08-12

AI Summary

To address the dual requirements of data efficiency and localization accuracy for camera pose estimation in retail environments, this paper proposes a relative pose regression method based on a Pose Autoencoder (PAE). Methodologically, we are the first to adapt PAE to relative pose estimation, jointly modeling scene spatial priors and visual features within an end-to-end framework that performs both relative pose regression and absolute pose refinement. We further design a lightweight relocalization strategy requiring no additional storage of images or poses. Our key contributions are: (1) a learnable pose prior representation mechanism; and (2) highly efficient pose refinement achieving state-of-the-art performance using only 30% of the training data. Evaluated on indoor benchmarks, our approach significantly improves localization accuracy while substantially reducing data collection costs for retail deployment.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Image and Video Retrieval
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Ad search and search for Web retail
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Accurate camera localization is crucial for modern retail environments, enabling enhanced customer experiences, streamlined inventory management, and autonomous operations. While Absolute Pose Regression (APR) from a single image offers a promising solution, approaches that incorporate visual and spatial scene priors tend to achieve higher accuracy. Camera Pose Auto-Encoders (PAEs) have recently been introduced to embed such priors into APR. In this work, we extend PAEs to the task of Relative Pose Regression (RPR) and propose a novel re-localization scheme that refines APR predictions using PAE-based RPR, without requiring additional storage of images or pose data. We first introduce PAE-based RPR and establish its effectiveness by comparing it with image-based RPR models of equivalent architectures. We then demonstrate that our refinement strategy, driven by a PAE-based RPR, enhances APR localization accuracy on indoor benchmarks. Notably, our method is shown to achieve competitive performance even when trained with only 30% of the data, substantially reducing the data collection burden for retail deployment. Our code and pre-trained models are available at: https://github.com/yolish/camera-pose-auto-encoders

Problem

• Enhancing camera localization accuracy for retail applications
• Reducing data collection burden with efficient pose regression
• Improving re-localization using auto-encoder priors without extra storage

Innovation

• Extends Pose Auto-Encoders to Relative Pose Regression
• Refines APR predictions using PAE-based RPR
• Achieves competitive performance with 30% data