A Probabilistic Framework for Visual Localization in Ambiguous Scenes

TL;DR

This paper introduces a probabilistic approach for visual localization in ambiguous scenes, modeling the full posterior distribution of camera poses using variational inference, which improves accuracy over existing methods.

Contribution

It presents a novel probabilistic framework that predicts the entire posterior distribution of camera poses, addressing ambiguity in scene localization.

Findings

Outperforms existing localization methods in ambiguous scenes

Predicts arbitrarily shaped posterior distributions of camera poses

Enables sampling from the predicted pose distribution

Abstract

Visual localization allows autonomous robots to relocalize when losing track of their pose by matching their current observation with past ones. However, ambiguous scenes pose a challenge for such systems, as repetitive structures can be viewed from many distinct, equally likely camera poses, which means it is not sufficient to produce a single best pose hypothesis. In this work, we propose a probabilistic framework that for a given image predicts the arbitrarily shaped posterior distribution of its camera pose. We do this via a novel formulation of camera pose regression using variational inference, which allows sampling from the predicted distribution. Our method outperforms existing methods on localization in ambiguous scenes. Code and data will be released at https://github.com/efreidun/vapor.