Combined Fitness Exercise Dataset

该数据集融合了来自InfiniteRep数据集的合成数据以及来自Kaggle和其他来源的真实世界视频，专注于运动分类。该数据集涵盖了受控和非受控环境，强调使用角度和坐标对于提高泛化能力的重要性。规模上，数据集包含多种运动类型，如深蹲、俯卧撑、肩部推举和二头肌弯举。任务方面，旨在进行运动分类和计数。

This dataset integrates synthetic data from the InfiniteRep dataset and real-world videos sourced from Kaggle and other resources, focusing on exercise classification. It covers both controlled and uncontrolled environments, and emphasizes the critical role of using angles and coordinates to improve generalization performance. The dataset includes a diverse range of exercise types such as squats, push-ups, shoulder presses, and bicep curls. Its intended downstream tasks are exercise classification and counting.

搜集汇总

背景与挑战

背景概述

该数据集是一个AI健身应用项目，利用计算机视觉和机器学习技术实现实时运动识别和计数。数据集包含预训练模型和工具脚本，支持多种健身动作的分类和计数，适用于健身辅助应用开发。