现代电子技术2024，Vol.47Issue(24)：143-147,5.DOI:10.16652/j.issn.1004-373x.2024.24.022
基于显著性特征的多视角人体动作图像识别研究
Research on multi-perspective human action image recognition based on significant features
廖民玲1
作者信息
- 1. 桂林电子科技大学,广西 桂林 541004
- 折叠
摘要
Abstract
In order to address the difficulty of recognizing human motion images with multiple angles and postures,a method of multi-perspective human motion image recognition based on salient features is researched.By using salient region detection module to process multi-perspective human motion images,a sequence of salient regions is obtained and concatenated into a multi-perspective salient region mosaic images.The convolutional feature map is extracted by means of VGG-Net network and input into CA(up and down attention)module.The region conducive to human action recognition is highlighted,and the probability value of human action category label is output to realize multi-perspective human action recognition.The experimental results show that the proposed method can effectively recognize multi-perspective human movements.By integrating salient region detection and CA modules,the application effectiveness of the method in multi-perspective human action recognition can be improved significantly.关键词
显著性特征/多视角/人体动作/图像识别/类别标签/CA模块/LSTM单元Key words
significant features/multi-perspective/human action/image recognition/category label/CA module/LSTM unit分类
信息技术与安全科学引用本文复制引用
廖民玲..基于显著性特征的多视角人体动作图像识别研究[J].现代电子技术,2024,47(24):143-147,5.
