ITSC 2025 Paper Abstract

Paper VP-VP.97

NICODEME, Claire (EDF), You, Borachhun (SNCF), Théo, Niemann (SNCF)

Pedestrian Trajectory Projection on Train Stations’ 2D Map for Optimized Station & Traffic Management

Scheduled for presentation during the Video Session "On-Demand Video Presentations" (VP-VP), Saturday, November 22, 2025, 08:00−18:00, On-Demand Platform

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on April 2, 2026

Keywords AI, Machine Learning for Real-time Traffic Flow Prediction and Management, Real-time Passenger Information and Service Optimization in Public Transportation, Deep Learning for Scene Understanding and Semantic Segmentation in Autonomous Vehicles

Abstract

For any public transportation system, the main goal is to provide travelers with a way to go from one location to another. The service (train operation and planning, shops, elevators, etc.) is organized around the attendance of stations and vehicles. Hence, estimating the number of people, how they are distributed, and how they move (i.e. their trajectories) is crucial for an efficient station and traffic management system. Additionally, it also helps improve service quality and passenger comfort. To this end, surveillance cameras are valuable sensors to capture data. On one hand, they are mandatory for security purposes, and many are already installed. On the other hand, they mimic the human eye. Images and videos provide the visual and intelligible information required by human agents. However, visualizing trajectories in the image reference frame is suboptimal: given the numerous cameras and the diversity of associated viewpoints, agents need a global view of passengers’ movements, not a camera-specific one. This study proposes a method for trajectory projection, from the reference frame of the image to the 2D map of the station. First, a pedestrian detection and tracking model counts people and extracts positions on video frames. Then, the trajectory is transformed to overcome perspective: it is plotted in the image plane, then in the camera reference frame and finally in the terrestrial frame. The result is a 2D map of stations with real-time attendance updates and extracted statistics.