节点文献

复杂场景下多重损失的步态识别系统的实现

Realization of Gait Recognition System with Multiple Losses in Complex Scenes

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 李长萍于大海郭豪李杰胡滨

【Author】 LI Changping;YU Dahai;GUO Hao;LI Jie;HU Bin;College of Computer Science and Technology, Guizhou University;

【通讯作者】 胡滨;

【机构】 贵州大学计算机科学与技术学院

【摘要】 在目前的高校教育制度下,我国的教育事业蓬勃发展,但是在学生课堂早退违纪问题的处理中仍存在不足之处,由于早退学生的过多遮挡以及背景环境较高的复杂程度,因此在识别学生的身份时很难同时兼顾速度和准确性。针对该问题,基于显著性目标检测U2Net网络和多重损失的GaitSet模型,结合步态识别相关技术,设计和实现了针对于高校课堂早退学生身份识别的系统DGRSML,为有效解决高校学生课堂早退违纪现象的身份识别问题提供了帮助。

【Abstract】 Under the present university education system, our education enterprise has been developing vigorously, but there are still some shortcomings in the process of the problem of early class leave disciplinary problems, because of the excessive cover of early students and the complexity of the environment, can not quickly and accurately identify the identity of the students. To solve this problem, based on significant object detection U2Net network and multiple loss GaitSet model, combined with gait recognition technology, the identification system for college students who leave class early is designed and implemented, which provides help to effectively solve the discipline problem of college students who leave class early.

【基金】 贵州大学大学生创新创业训练计划项目“基于步态特征的课堂早退学生身份识别系统设计与实现”(贵大省创字(2021)050);贵州大学引进人才科研项目(贵大人基合字(2019)58号);贵州省自然科学基金项目(黔科合基础[2020]1Y261)
  • 【分类号】TP391.41
  • 【下载频次】11
节点文献中: 

本文链接的文献网络图示:

本文的引文网络