RelMobNet: End-to-end relative camera pose estimation using a robust two-stage training

TL;DR

This paper introduces RelMobNet, an end-to-end siamese network for relative camera pose estimation that improves accuracy and generalization through a novel two-stage training process, outperforming existing CNN-based methods.

Contribution

The paper proposes a new two-stage training approach for a siamese network that enhances translation accuracy and generalization in relative pose estimation without relying on camera parameters.

Findings

Improves translation vector estimation by up to 52.27% on certain scenes.

Demonstrates better generalization across different scene styles using GAN-based augmentation.

Provides qualitative analysis of epipolar lines aligning with ground truth poses.

Abstract

Relative camera pose estimation, i.e. estimating the translation and rotation vectors using a pair of images taken in different locations, is an important part of systems in augmented reality and robotics. In this paper, we present an end-to-end relative camera pose estimation network using a siamese architecture that is independent of camera parameters. The network is trained using the Cambridge Landmarks data with four individual scene datasets and a dataset combining the four scenes. To improve generalization, we propose a novel two-stage training that alleviates the need of a hyperparameter to balance the translation and rotation loss scale. The proposed method is compared with one-stage training CNN-based methods such as RPNet and RCPNet and demonstrate that the proposed model improves translation vector estimation by 16.11%, 28.88%, and 52.27% on the Kings College, Old Hospital, and St Marys Church scenes, respectively. For proving texture invariance, we investigate the generalization of the proposed method augmenting the datasets to different scene styles, as ablation studies, using generative adversarial networks. Also, we present a qualitative assessment of epipolar lines of our network predictions and ground truth poses.