Improving Nighttime Retrieval-Based Localization
This addresses a critical bottleneck for outdoor computer vision systems that need to operate reliably across different lighting conditions, representing a strong specific gain in a domain-specific area.
The paper tackles the problem of visual localization under strong appearance changes between daytime and nighttime, proposing a method that improves localization by almost 300% compared to VLAD-based methods on nighttime data.
Outdoor visual localization is a crucial component to many computer vision systems. We propose an approach to localization from images that is designed to explicitly handle the strong variations in appearance happening between daytime and nighttime. As revealed by recent long-term localization benchmarks, both traditional feature-based and retrieval-based approaches still struggle to handle such changes. Our novel localization method combines a state-of-the-art image retrieval architecture with condition-specific sub-networks allowing the computation of global image descriptors that are explicitly dependent of the capturing conditions. We show that our approach improves localization by a factor of almost 300\% compared to the popular VLAD-based methods on nighttime localization.