Data is an important aspect of modern deep learning research, particularly in Place Recognition, which plays a pivotal role in various applications such as robotics, augmented reality, and autonomous navigation. In this paper, we introduce the ITLP-Campus dataset captured by a mobile robot equipped with front and back cameras and a LiDAR sensor. The dataset encompasses diverse outdoor and indoor environments within a university campus setting. Spanning different times of day, seasons, and weather conditions, ITLP-Campus offers a rich and varied set of scenes for analysis. The dataset includes unique features such as strategically placed ArUco markers along routes, automatically generated semantic masks and textual descriptions of scenes. Moreover, indoor tracks are annotated with scene text, enhancing the dataset’s utility for tasks involving text recognition and understanding. We provide detailed insights into the dataset’s acquisition process, annotation procedures, and potential applications. Additionally, we conduct extensive experiments by testing popular Place Recognition methods on ITLP-Campus, demonstrating its effectiveness and versatility for advancing research in multimodal semantic place recognition and related fields.
Alexander Melekhin, Vitaly Bezuglyj, Dmitry Yudin, Ilia Petryashin, Kirill Muravyev, Sergey Linok and Aleksandr Panov. TLP-Campus: A Dataset for Multimodal Semantic Place Recognition // Proceedings of the Eighth International Scientific Conference “Intelligent Information Technologies for Industry” (IITI’24), Volume 1. IITI 2024. Lecture Notes in Networks and Systems. Vol 1209. Pp. 185–195.