Scholarlink

Driver drowsiness detection is vital for road safety, addressing challenges like diverse lighting, resolutions, and sensor types. This study integrates IoT with a fine-tuned ResNet50 model for feature extraction and classification on multi-resolution eye images. Selective layer unfreezing, coupled with Early Stopping and ReduceLROnPlateau, improves training stability and generalization. Comparative analysis shows ResNet50 outperforms architectures like DenseNet, MobileNet, and ANN, achieving high accuracy and robustness. The system's effectiveness in real-time IoT-enabled environments highlights its potential to enhance driver alertness detection and road safety under varied conditions.

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

IoT and Deep Learning Approaches for Enhanced Driver Drowsiness Detection Using Multi-Resolution Eye Image Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Vibha Kulkarni,
  • L. Lakshmi Prasanna Kumar,
  • Majeti Venkata Sireesha

摘要

Driver drowsiness detection is vital for road safety, addressing challenges like diverse lighting, resolutions, and sensor types. This study integrates IoT with a fine-tuned ResNet50 model for feature extraction and classification on multi-resolution eye images. Selective layer unfreezing, coupled with Early Stopping and ReduceLROnPlateau, improves training stability and generalization. Comparative analysis shows ResNet50 outperforms architectures like DenseNet, MobileNet, and ANN, achieving high accuracy and robustness. The system's effectiveness in real-time IoT-enabled environments highlights its potential to enhance driver alertness detection and road safety under varied conditions.