We are pleased to announce that a paper led by Riki Ukyo, who completed his Ph.D. in the 2024 academic year, titled "Robust Pedestrian Tracking With Severe Occlusions in Public Spaces Using 3D Point Clouds," has been accepted for publication in the IEEE Transactions on Intelligent Transportation Systems, a renowned international journal in the field of transportation systems.
- R. Ukyo, T. Amano, H. Rizk, H. Yamaguchi and T. Moriya, "Robust Pedestrian Tracking With Severe Occlusions in Public Spaces Using 3D Point Clouds," in IEEE Transactions on Intelligent Transportation Systems, doi: 10.1109/TITS.2025.3547418.
This research focuses on technology for accurately tracking pedestrian flow in public spaces such as train stations, commercial facilities, and event venues, while giving careful consideration to privacy. Traditional methods using camera footage for monitoring pedestrian movement have raised significant privacy issues because they can easily identify individuals. Therefore, this study utilizes "3D point cloud data" obtained from 3D sensors (like LiDAR or depth cameras), which makes identifying individuals difficult.
However, a challenge arises in crowded environments where "occlusion" frequently occurs – situations where people block one another from the sensor's view, making accurate tracking difficult. This is particularly problematic when pedestrians with strollers or large luggage are hidden behind others. This paper proposes a novel method that, without relying on AI, can accurately track people's positions by compensating for these hidden parts even when such occlusions occur. This technology enables more robust tracking of individuals without interruption, even in densely crowded situations.
The effectiveness of the proposed method was validated not only in laboratory settings but also using data collected from a real-world shopping mall entrance and during a festival in Kita-ku, Osaka City, confirming its high accuracy.
This achievement is expected to contribute to the realization of safer and more comfortable public spaces and the development of new, privacy-preserving pedestrian flow analysis technologies.
