Calisthenics Skills Temporal Video Segmentation Dataset
收藏资源简介:
该数据集由卡塔尼亚大学数学与计算机科学系的研究人员创建,包含839个视频,记录了运动员执行9种静态体操技能的情况。每个视频都经过人工标注,确定了每个技能的开始和结束时间。数据集旨在帮助运动员在训练中识别和量化技能的持续时间,并为裁判在比赛中提供辅助。数据集的视频来自不同的来源,包括社交网络和专门录制的视频,以确保视频的真实性和自然性。数据集还包括使用OpenPose技术提取的运动员关节坐标。数据集分为训练集和测试集,其中训练集包含80%的视频,测试集包含20%的视频。
This dataset was created by researchers from the Department of Mathematics and Computer Science, University of Catania. It consists of 839 videos capturing athletes performing 9 types of static gymnastic skills. Each video has been manually annotated to determine the start and end timestamps of each skill. The dataset aims to assist athletes in identifying and quantifying the duration of their skills during training, and provide support for referees during competitions. The videos are sourced from various channels including social networks and specially recorded footage to ensure the authenticity and naturalness of the content. The dataset also includes joint coordinates of athletes extracted using OpenPose technology. It is split into training and test sets, with the training set accounting for 80% of the total videos and the test set making up the remaining 20%.
Calisthenics Skills Temporal Video Segmentation 数据集概述
数据集基本信息
- 数据集名称: Calisthenics Skills Temporal Video Segmentation
- 相关论文:
Finocchiaro, A.; Farinella, G. and Furnari, A. (2024). Calisthenics Skills Temporal Video Segmentation. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP, ISBN 978-989-758-679-8, ISSN 2184-4321, pages 182-190.
数据集内容
- 数据目录:
/data/keypoints_dataset - 数据类型: 关键点数据集
相关代码
- 代码目录:
/src- 推理脚本:
codec.py: 编码和解码标签inference.py: 主推理脚本,测试整个流程openpose_script.py: 从视频中提取关键点
- 模型:
mlp.py: 定义多层感知机(MLP)架构
- 时间分割:
furnari2018.py: 概率算法实现heuristic.py: 启发式算法实现
- 推理脚本:
使用要求
- 依赖库: 安装
requirements.txt中的库 - 额外工具: 需要安装OpenPose




