Reproduce “An Indoor Crowd Movement Trajectory Benchmark Dataset” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ying Zhao, Xin Zhao, Siming Chen, Zhuo Zhang, Xin Huang
This paper introduces the Indoor Crowd Movement Trajectory Dataset (ICMTD-2019), a publicly available benchmark dataset generated through a program-driven simulation of over 5,000 participants at a three-day conference. The dataset includes realistic movement patterns and annotated anomalous events, and its utility was validated through its use in the ChinaVis Data Challenge 2019, where it served as a robust evaluation tool for indoor crowd analysis methods.
Key insights
- A program-driven simulation approach can effectively generate realistic and diverse indoor crowd movement trajectories, overcoming challenges of real-world data collection like privacy and technological limitations.
- Designing datasets with principles of realism, universality, diversity, and challenge allows for the creation of robust benchmarks for evaluating analytical methods.
- The inclusion of both normal and anomalous events, with ground truth annotations, is crucial for developing and testing anomaly detection algorithms in indoor environments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.