摘要
采用“CNN+LSTM”组合深度神经网络结构,文章研究并设计了一个新型实时监控打架视频动态预测系统,不仅能通过CNN提取动作行为的图像特征,还可以通过LSTM预测是否存在打架行为,同时使用OpenCV进行实时视频监控及图像处理,最后通过PyQt5界面显示实时行为视频与预测结果。
Using the“CNN+LSTM”combination deep neural network structure,this article studies and designs a new real-time monitoring and fighting video dynamic prediction system.It can not only extract image features of action behavior through CNN,but also predict whether there is fighting behavior through LSTM.At the same time,OpenCV is used for real-time video monitoring and image processing.Finally,the real-time behavior video and prediction results are visualized through the PyQt5 interface.
作者
王蕊
WANG Rui(China Mobile Communications Group Guangdong Co.,Ltd.,Guangzhou 510627,China)
出处
《计算机应用文摘》
2024年第4期108-110,113,共4页
Chinese Journal of Computer Application
