Scholarlink

<p>The efficient utilization of educational resources in academic settings is a critical concern. This study introduces an integrated facial recognition system using established deep learning techniques to support efficient utilization of university classrooms through real-time monitoring and attendance tracking. Objectives include: (1) combining MobileNet-based feature extraction with an SSA-SVM classification model for enhanced classroom management and resource allocation; (2) assessing performance in recognition accuracy and real-time processing. The model employs a depthwise separable convolutional network with an inverted residual module for precise feature extraction, paired with SSA-SVM for categorization. This supports real-time monitoring and resource optimization. Tested across Henan Province institutions, it achieved a 3.47% increase in face detection accuracy and 7.05% in recognition rate over baselines, reaching 98.13% accuracy. In classrooms, it delivered 93.61% accuracy at 125 fps, surpassing baseline methods in efficiency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An integrated facial recognition system for classroom resource optimization using MobileNet and SSA-SVM

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月29日

查看全文
  • Kaiyuan Fan

摘要

The efficient utilization of educational resources in academic settings is a critical concern. This study introduces an integrated facial recognition system using established deep learning techniques to support efficient utilization of university classrooms through real-time monitoring and attendance tracking. Objectives include: (1) combining MobileNet-based feature extraction with an SSA-SVM classification model for enhanced classroom management and resource allocation; (2) assessing performance in recognition accuracy and real-time processing. The model employs a depthwise separable convolutional network with an inverted residual module for precise feature extraction, paired with SSA-SVM for categorization. This supports real-time monitoring and resource optimization. Tested across Henan Province institutions, it achieved a 3.47% increase in face detection accuracy and 7.05% in recognition rate over baselines, reaching 98.13% accuracy. In classrooms, it delivered 93.61% accuracy at 125 fps, surpassing baseline methods in efficiency.