FLAG3D
收藏资源简介:
FLAG3D是一个大规模的3D健身活动数据集,由清华大学创建,包含180000个序列,涵盖60种日常健身活动。数据集通过先进的动作捕捉系统、渲染软件和智能手机在自然环境中捕捉,提供精确密集的3D人体姿态和详细的语言指导。该数据集旨在支持跨领域的人类动作识别、动态人体网格恢复和语言引导的人类动作生成等研究,为健身活动分析提供丰富的资源和挑战。
FLAG3D is a large-scale 3D fitness activity dataset created by Tsinghua University, containing 180,000 sequences covering 60 types of daily fitness activities. Captured in natural environments using advanced motion capture systems, rendering software and smartphones, the dataset provides precise and dense 3D human poses and detailed linguistic guidance. It aims to support research in cross-domain human action recognition, dynamic human mesh recovery, language-guided human action generation and other related fields, offering abundant resources and challenges for fitness activity analysis.

- 1FLAG3D: A 3D Fitness Activity Dataset with Language Instruction清华大学 · 2023年



