WorldExpo'10 is a large-scale crowd counting and density estimation dataset collected from surveillance camera footage at the 2010 Shanghai World Expo, comprising hundreds of video sequences captured across multiple scenes with varying crowd densities and viewpoints. It is significant to the field because it provides a challenging, real-world benchmark with perspective map annotations and region of interest labels that enable evaluation of crowd counting models under diverse environmental conditions and severe perspective distortion. The dataset is commonly used in its standard split form for training and testing CNN-based crowd counting methods, and its multi-scene structure makes it particularly valuable for assessing model generalization across heterogeneous crowd scenarios.
Source Papers
- A survey of recent advances in CNN-based single image crowd counting and density estimation ↗ — A survey of recent advances in CNN-based single image crowd
- Data-driven Crowd Modeling Techniques: A Survey ↗ — Data-driven Crowd Modeling Techniques: A Survey
- Recent trends in crowd analysis: A review ↗ — Recent trends in crowd analysis: A review