Scholarlink

Eye blink detection is a computer vision technology that involves identifying and determining landmarks on a face. Blink detection is verified by examining the eye-aspect ratio (EAR). The eAuthenticate is a real-time facial recognition technique for detecting eye blinking. Facial landmarks are detected by a combination of histogram-oriented gradients (HOG) and shape prediction, which is helpful for feature descriptors. It uses an ensemble of regression trees, which is trained on a large number of datasets. For each video frame, all landmarks are detected on a face. Among them, the proposed technique will find eye landmark locations and calculate the EAR by taking the ratio of the vertical distance of an eyelid to the double horizontal distance of an eyelid. From performance analysis, we can infer that the proposed model is a significantly detected eye landmark for real-time faces with high accuracy.

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

eAuthenticate: Enhancing Security with Machine Learning Based Real-Time Open-World Anti-spoofing Method

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Om Delvadia,
  • Jitendra Parmar,
  • Pratik Gupta,
  • Kuruba Lakshmi Harsha Vardhan,
  • Dhruv Gandhi,
  • Atul Kumar Verma

摘要

Eye blink detection is a computer vision technology that involves identifying and determining landmarks on a face. Blink detection is verified by examining the eye-aspect ratio (EAR). The eAuthenticate is a real-time facial recognition technique for detecting eye blinking. Facial landmarks are detected by a combination of histogram-oriented gradients (HOG) and shape prediction, which is helpful for feature descriptors. It uses an ensemble of regression trees, which is trained on a large number of datasets. For each video frame, all landmarks are detected on a face. Among them, the proposed technique will find eye landmark locations and calculate the EAR by taking the ratio of the vertical distance of an eyelid to the double horizontal distance of an eyelid. From performance analysis, we can infer that the proposed model is a significantly detected eye landmark for real-time faces with high accuracy.