HiEve
收藏资源简介:
HiEve是一个大规模的数据集,专注于复杂事件中的人类中心视频分析。由上海交通大学电子工程系创建,包含32个视频序列,主要用于理解人类动作、姿势和行为。数据集包含超过100万个姿势和56,000多个动作实例,适用于多对象跟踪、姿势估计和动作识别等任务。HiEve特别关注于拥挤和复杂事件场景,旨在通过其多样化的标注提升视频理解技术。
HiEve is a large-scale dataset dedicated to human-centric video analysis in complex event scenarios. It was created by the Department of Electronic Engineering, Shanghai Jiao Tong University, and comprises 32 video sequences, primarily intended for understanding human actions, postures and behaviors. The dataset contains over 1 million pose annotations and more than 56,000 action instances, and is suitable for tasks including multi-object tracking, pose estimation and action recognition. HiEve specifically focuses on crowded and complex event scenarios, with the goal of advancing video understanding technologies via its diverse annotations.

- 1Human in Events: A Large-Scale Benchmark for Human-centric Video Analysis in Complex Events上海交通大学电子工程系 · 2023年



