2020/08/05 by Johanna Wald, Wald, Johanna, Torsten Sattler +7 · 5 citations
Computer Science · Engineering · #Advanced Image and Video Retrieval Techniques #Advanced Vision and Imaging #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Robotics and Sensor-Based Localization
paper · pdf · doi:10.48550/arxiv.2008.02004
openalex publication_date 2020/08/05 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Long-term camera re-localization is an important task with numerous computer\nvision and robotics applications. Whilst various outdoor benchmarks exist that\ntarget lighting, weather and seasonal changes, far less attention has been paid\nto appearance changes that occur indoors. This has led to a mismatch between\npopular indoor benchmarks, which focus on static scenes, and indoor\nenvironments that are of interest for many real-world applications. In this\npaper, we adapt 3RScan - a recently introduced indoor RGB-D dataset designed\nfor object instance re-localization - to create RIO10, a new long-term camera\nre-localization benchmark focused on indoor scenes. We propose new metrics for\nevaluating camera re-localization and explore how state-of-the-art camera\nre-localizers perform according to these metrics. We also examine in detail how\ndifferent types of scene change affect the performance of different methods,\nbased on novel ways of detecting such changes in a given RGB-D frame. Our\nresults clearly show that long-term indoor re-localization is an unsolved\nproblem. Our benchmark and tools are publicly available at\nwaldjohannau.github.io/RIO10\n