基于双流网络的人体动作识别
Human Motion Recognition Based on Two Stream Network
张艳 1张博 1肖文琛1
作者信息
- 1. 北华航天工业学院计算机学院,河北廊坊 065000
- 折叠
摘要
人体动作识别是视觉理解领域里的一个重要课题,涉及计算机视觉、人工智能等多个领域,在人机交互、视频监控,智慧医疗等方面有着广阔的应用前景.本文首先主要从传统的人体动作识别和基于深度学习的人体动作识别算法等方面进行分析探讨,然后阐述了本文提出的双流网络动作识别算法的主要思想,通过实验证明了算法的有效性.
Abstract
Human motion recognition is an important topic in the field of visual understanding,involving computer vision,artificial intelligence and other fields.It has broad application prospects in human-computer interaction,video surveillance,intelligent medical treatment and other aspects.This paper first analyzes and discusses the traditional human motion recognition and the human motion recognition algorithm based on deep learning,and then expounds the main idea of the two-stream network motion recognition algorithm,which proves its effectiveness through experiments.
关键词
双流网络/动作识别/深度学习/神经网络Key words
two-stream network/motion recognition/deep learning/neural network引用本文复制引用
基金项目
河北省高等学校科学技术研究项目(ZC2021006)
廊坊市科技局项目(2022011003)
北华航天工业学院研究生创新基金项目(YKY202238)
出版年
2024
NETL
NSTL
万方数据