Coarse-to-Fine Multi-Scene Pose Regression with Transformers

TL;DR

This paper introduces a Transformer-based approach for multi-scene camera pose regression, enabling the model to focus on relevant features and outperform existing methods on standard benchmarks.

Contribution

The work presents a novel Transformer architecture with mixed classification-regression for multi-scene pose estimation, improving accuracy over prior models.

Findings

Outperforms state-of-the-art single-scene regressors

Effective multi-scene localization on benchmark datasets

Transformer-based architecture enhances feature focus

Abstract

Absolute camera pose regressors estimate the position and orientation of a camera given the captured image alone. Typically, a convolutional backbone with a multi-layer perceptron (MLP) head is trained using images and pose labels to embed a single reference scene at a time. Recently, this scheme was extended to learn multiple scenes by replacing the MLP head with a set of fully connected layers. In this work, we propose to learn multi-scene absolute camera pose regression with Transformers, where encoders are used to aggregate activation maps with self-attention and decoders transform latent features and scenes encoding into pose predictions. This allows our model to focus on general features that are informative for localization, while embedding multiple scenes in parallel. We extend our previous MS-Transformer approach \cite{shavit2021learning} by introducing a mixed classification-regression architecture that improves the localization accuracy. Our method is evaluated on commonly benchmark indoor and outdoor datasets and has been shown to exceed both multi-scene and state-of-the-art single-scene absolute pose regressors.