Scholarlink

<p>The demand for non-contact biometric methods, such as human gait recognition, has increased significantly in recent years, driven by technological advancements and growing global security concerns, particularly during the COVID-19 pandemic. Gait recognition is valuable for various applications, including criminal investigations and virtual reality, but its accuracy can be influenced by factors such as clothing variations, walking conditions, and the carrying of loads. This study proposes an efficient approach to address these challenges by employing a convolutional neural network (CNN) for feature extraction and classification. Evaluations on the CASIA-B gait dataset, focusing on three specific angles (54°, 90°, and 180°), demonstrate impressive validation accuracy rates of 99.99%, 99.98%, and 99.98%, respectively, along with low loss values of 0.0016, 0.0023, and 0.0025. These results underscore the effectiveness of the proposed method in improving human gait recognition under diverse conditions.</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

A Convolutional Neural Network Approach for Gait Recognition Using the CASIA B Dataset: A Comprehensive Evaluation

  • 期刊论文

  • 发表日期：

    2025年7月22日

查看全文
  • Junainah Abd Hamid,
  • J. Gowrishankar,
  • Shivakrishna Dasi,
  • S. Srinadh Raju,
  • Mandeep Kaur Chohan,
  • Deeksha Verma,
  • Ahmed Alkhayyat,
  • Rajesh Singh

摘要

The demand for non-contact biometric methods, such as human gait recognition, has increased significantly in recent years, driven by technological advancements and growing global security concerns, particularly during the COVID-19 pandemic. Gait recognition is valuable for various applications, including criminal investigations and virtual reality, but its accuracy can be influenced by factors such as clothing variations, walking conditions, and the carrying of loads. This study proposes an efficient approach to address these challenges by employing a convolutional neural network (CNN) for feature extraction and classification. Evaluations on the CASIA-B gait dataset, focusing on three specific angles (54°, 90°, and 180°), demonstrate impressive validation accuracy rates of 99.99%, 99.98%, and 99.98%, respectively, along with low loss values of 0.0016, 0.0023, and 0.0025. These results underscore the effectiveness of the proposed method in improving human gait recognition under diverse conditions.