Scholarlink

Biometric-based gender prediction, mainly from eye traits such as Iris patterns, has huge research applications, i.e. security, personalized marketing, and health care. In this paper, we propose IGC which is an Iris-based gender classification system by using stable and unreplaceable texture feature of Iris, which shows the discriminator to predict a person’s gender with high accuracy. In this study, we investigate the utility of the segmented U-Net and seek to fine-grain the Iris features for gender classification. Our method shows better accuracy (92%), precision (90%), recall (85%), and AUC of 0.93. These results indicate that U-Net is capable of separating male and female Irises, which can be utilized for stronger gender-enrolment systems. In the analysis of results, it is evident that our model achieves a trade-off accuracy which plays a significant role in real-world applications such as biometric authentication and security systems.

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

Gender Prediction Using Iris Biometrics: A U-Net-Based Deep Learning Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Rashmi Sharma,
  • Nitin Rakesh,
  • Monali Gulhane,
  • Priya Dasarwar,
  • Ankita Tiwari,
  • Saurav Dixit

摘要

Biometric-based gender prediction, mainly from eye traits such as Iris patterns, has huge research applications, i.e. security, personalized marketing, and health care. In this paper, we propose IGC which is an Iris-based gender classification system by using stable and unreplaceable texture feature of Iris, which shows the discriminator to predict a person’s gender with high accuracy. In this study, we investigate the utility of the segmented U-Net and seek to fine-grain the Iris features for gender classification. Our method shows better accuracy (92%), precision (90%), recall (85%), and AUC of 0.93. These results indicate that U-Net is capable of separating male and female Irises, which can be utilized for stronger gender-enrolment systems. In the analysis of results, it is evident that our model achieves a trade-off accuracy which plays a significant role in real-world applications such as biometric authentication and security systems.