AtLoc: Attention Guided Camera Localization

TL;DR

This paper introduces AtLoc, a novel attention-guided approach for camera localization that enhances robustness and accuracy using single images by focusing on geometrically stable features, outperforming existing methods.

Contribution

The work demonstrates that attention mechanisms can improve single-image camera localization by emphasizing robust features, achieving state-of-the-art results and effectively rejecting dynamic objects.

Findings

State-of-the-art performance on benchmark datasets

Effective rejection of dynamic objects via saliency maps

Robust camera pose estimation with single images

Abstract

Deep learning has achieved impressive results in camera localization, but current single-image techniques typically suffer from a lack of robustness, leading to large outliers. To some extent, this has been tackled by sequential (multi-images) or geometry constraint approaches, which can learn to reject dynamic objects and illumination conditions to achieve better performance. In this work, we show that attention can be used to force the network to focus on more geometrically robust objects and features, achieving state-of-the-art performance in common benchmark, even if using only a single image as input. Extensive experimental evidence is provided through public indoor and outdoor datasets. Through visualization of the saliency maps, we demonstrate how the network learns to reject dynamic objects, yielding superior global camera pose regression performance. The source code is avaliable at https://github.com/BingCS/AtLoc.