Embodied Crowd Counting Dataset (ECCD)
收藏资源简介:
Embodied Crowd Counting Dataset (ECCD)是由哈尔滨工业大学(深圳)和香港城市大学联合创建的一个大型户外人群计数数据集。该数据集支持60个不同的大型户外场景,每个场景覆盖面积可达40,000平方米,目标数量可达15,000个。数据集采用泊松点过程来模拟真实的人群分布情况。ECCD旨在为人群计数研究提供一个具有交互能力的测试环境,支持大规模户外场景和大量目标的多样性分布,可用来评估和设计新型的人群计数算法。
Embodied Crowd Counting Dataset (ECCD) was jointly developed by Harbin Institute of Technology (Shenzhen) and City University of Hong Kong as a large-scale outdoor crowd counting dataset. It includes 60 distinct large-scale outdoor scenarios, each spanning up to 40,000 square meters and accommodating up to 15,000 targets. The dataset utilizes Poisson point processes to simulate realistic crowd distributions. ECCD aims to provide an interactive test environment for crowd counting research, supporting large-scale outdoor scenarios and diverse distributions of massive targets, and can be employed to evaluate and design novel crowd counting algorithms.

- 1Embodied Crowd Counting哈尔滨工业大学(深圳), 香港城市大学 · 2025年



