Constructing “RTRI-60”: A dataset of security camera recordings with controlled annotations for anomaly detection within railway environments
The Japanese Journal of Cognitive Psychology · 2026
DOI: 10.5265/jcogpsy.2604 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Japanese Journal of Cognitive Psychology, 2026. Verfasst von Nao Hanyu und 3 weiteren Autor:innen.
Abstract (Originalsprache)
A key priority for railway operators is to ensure safety. One particularly urgent challenge is to develop technologies capable of detecting abnormal situations that threaten passengers as early and reliably as possible. However, developing such detection systems for railway environments faces two main obstacles. The first is a lack of video datasets tailored to Japanese railway environments to use in creating and training detection algorithms. The second barrier is a lack of understanding about how continuous anomaly-judgment rates from multiple monitors vary for environments where diverse incidents can occur, which as stabbings or suicide attempts. This makes it particular problematic to know which cognitive characteristics should be prioritized in designing effective systems. To address these challenges, this study constructs RTRI-60, a dataset of security camera footage from railway environments that has been annotated by 25 annotators. This dataset can be utilized not only for developing and evaluating anomaly-detection algorithms but also as behavioral data for empirically assessing the cognitive characteristics of surveillance personnel.
Zitieren
APA
Hanyu, N., Takahashi, H., Kuwabara, S., & Nagamine, N. (2026). Constructing “RTRI-60”: A dataset of security camera recordings with controlled annotations for anomaly detection within railway environments. The Japanese Journal of Cognitive Psychology. https://doi.org/10.5265/jcogpsy.2604
Vancouver
Hanyu N, Takahashi H, Kuwabara S, Nagamine N. Constructing “RTRI-60”: A dataset of security camera recordings with controlled annotations for anomaly detection within railway environments. The Japanese Journal of Cognitive Psychology. 2026. doi:10.5265/jcogpsy.2604
BibTeX
@article{Hanyu2026,
author = {Hanyu, Nao and Takahashi, Hiroyuki and Kuwabara, Shohei and Nagamine, Nozomi},
title = {Constructing “RTRI-60”: A dataset of security camera recordings with controlled annotations for anomaly detection within railway environments},
journal = {The Japanese Journal of Cognitive Psychology},
year = {2026},
doi = {10.5265/jcogpsy.2604}
}
EUVIMED