Crowd_movement_time_series
DOI10.5281/zenodo.14899389Zenodo14899389MaRDI QIDQ6720285FDOQ6720285
Dataset published at Zenodo repository.
Petr Dolezel, Daniel Honc, Dominik Štursa, Dušan Kopecký
Publication date: 17 January 2025
Copyright license: Creative Commons Attribution 4.0 International
This dataset consists of JSON files that store information about the location and state of moving agents (representing people) in a defined space. Each JSON file corresponds to a specific time frame in a sequence, allowing for tracking agents across multiple frames. The dataset is structured to facilitate motion analysis, behavior recognition, and other spatial-temporal studies of human movement. Each file follows the naming pattern frame_000001.json, frame_000002.json, and so on, where frame_xxxxxx.json represents a specific timestamped frame in the sequence, and the numbering increments sequentially for each successive frame. Each JSON file contains a list of agents present in that frame, with each agents data represented as an object with several attributes. The Name attribute serves as a unique identifier for the agent, such as BP_Agent_C_0 or BP_Agent_C_1, and remains consistent across frames, allowing tracking over time. The Size attribute is a floating-point value representing the physical size of the agent in the simulation, while the Speed attribute indicates the agents movement speed. The State attribute defines the behavioral state of the agent, which can be Idle if the agent is stationary, Walking if moving at a normal pace, Running if moving at an accelerated pace, or Fighting if engaged in combat or conflict. The position of each agent is described using two sets of coordinates: GroundWS (World Space Position - Ground Level) and HeadWS (World Space Position - Head Level). Each of these contains X, Y, and Z values representing the agent's location at ground level and head level, respectively. Each frame may contain a varying number of agents, depending on their presence in the scene. The consistency of the Name attribute across frames allows for temporal tracking of individual agents movement and behavior over time. This dataset is suitable for motion tracking and behavioral analysis, crowd simulation studies, AI-based movement prediction models, and augmented reality or game development applications. By leveraging the structured naming conventions and JSON data format, researchers and developers can efficiently process, analyze, and visualize human movement patterns over time.
This page was built for dataset: Crowd_movement_time_series
