Adversarial Learning of Hard Positives for Place Recognition

TL;DR

This paper introduces an adversarial approach to generate hard positive examples for training image retrieval networks, enhancing robustness and accuracy in place recognition tasks.

Contribution

It proposes a novel adversarial method that learns local and global augmentation policies to produce challenging training examples, improving feature robustness beyond traditional augmentation techniques.

Findings

Achieves state-of-the-art recalls on Pitts250 and Tokyo 24/7 benchmarks.

Outperforms recent methods on rOxford and rParis datasets.

Enhances robustness of image retrieval features to variations.

Abstract

Image retrieval methods for place recognition learn global image descriptors that are used for fetching geo-tagged images at inference time. Recent works have suggested employing weak and self-supervision for mining hard positives and hard negatives in order to improve localization accuracy and robustness to visibility changes (e.g. in illumination or view point). However, generating hard positives, which is essential for obtaining robustness, is still limited to hard-coded or global augmentations. In this work we propose an adversarial method to guide the creation of hard positives for training image retrieval networks. Our method learns local and global augmentation policies which will increase the training loss, while the image retrieval network is forced to learn more powerful features for discriminating increasingly difficult examples. This approach allows the image retrieval network to generalize beyond the hard examples presented in the data and learn features that are robust to a wide range of variations. Our method achieves state-of-the-art recalls on the Pitts250 and Tokyo 24/7 benchmarks and outperforms recent image retrieval methods on the rOxford and rParis datasets by a noticeable margin.