Scholarlink

<p>In recent years, distracted driving is considered as one of the most significant reasons for traffic accidents. Therefore, the ability to detect inattentive drivers is essential to create safe intelligent transportation. Numerous research efforts have been made to solve the issue via different techniques but are infeasible for mass production. Diverse research has been performed using deep learning approaches to detect distracted drivers but due to limited efficiency and accuracy, the effectiveness of these systems remains a challenging one. These challenges are addressed by proposing a novel Stacked Attention Bidirectional Long Short-term based Greylag Goose Search algorithm. Residual Network is used to capture the driver distraction visuals, where the driver’s behavior is tracked using the stacked long short-term memory. In addition, an attention mechanism is utilized to focus the specific parts of video frames. In this paper, hyperparameter optimization is carried out by utilizing the Greylag Goose optimization with an initial search strategy that enhances the effectiveness of the proposed model. The Frame Level Driver Sleep Detection (FL3D) dataset and the American University in Cairo (AUC) Distracted Driver dataset are used to validate the proposed model. Different performance metrics including the mean estimated error, and root mean square error are used to evaluate the performance of the proposed model when compared to the existing models. Therefore, the experimental results evaluate the performance of the proposed model in detecting driver distraction and fatigue. The performance of the proposed method is compared with other methods such as E2DR, OWIPA, HSDDD, Fuzzy-macro- LSTM, and AT-Bi-LSTM. The proposed method achieved 98.71% accuracy for the FL3D dataset, 98.53% accuracy for the AUC distracted driver dataset, 98.32% precision for the FL3D dataset, and 98.10% precision for the AUC distracted driver dataset. Overall, the experimental results demonstrated that the proposed model outperformed all existing models in predicting both driver sleepiness or fatigue level and distraction level.</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

Social network analysis in driver monitoring systems: understanding the impact of predictive models on driver safety

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年4月6日

查看全文
  • S. Madhumita,
  • A. Jeyasekar

摘要

In recent years, distracted driving is considered as one of the most significant reasons for traffic accidents. Therefore, the ability to detect inattentive drivers is essential to create safe intelligent transportation. Numerous research efforts have been made to solve the issue via different techniques but are infeasible for mass production. Diverse research has been performed using deep learning approaches to detect distracted drivers but due to limited efficiency and accuracy, the effectiveness of these systems remains a challenging one. These challenges are addressed by proposing a novel Stacked Attention Bidirectional Long Short-term based Greylag Goose Search algorithm. Residual Network is used to capture the driver distraction visuals, where the driver’s behavior is tracked using the stacked long short-term memory. In addition, an attention mechanism is utilized to focus the specific parts of video frames. In this paper, hyperparameter optimization is carried out by utilizing the Greylag Goose optimization with an initial search strategy that enhances the effectiveness of the proposed model. The Frame Level Driver Sleep Detection (FL3D) dataset and the American University in Cairo (AUC) Distracted Driver dataset are used to validate the proposed model. Different performance metrics including the mean estimated error, and root mean square error are used to evaluate the performance of the proposed model when compared to the existing models. Therefore, the experimental results evaluate the performance of the proposed model in detecting driver distraction and fatigue. The performance of the proposed method is compared with other methods such as E2DR, OWIPA, HSDDD, Fuzzy-macro- LSTM, and AT-Bi-LSTM. The proposed method achieved 98.71% accuracy for the FL3D dataset, 98.53% accuracy for the AUC distracted driver dataset, 98.32% precision for the FL3D dataset, and 98.10% precision for the AUC distracted driver dataset. Overall, the experimental results demonstrated that the proposed model outperformed all existing models in predicting both driver sleepiness or fatigue level and distraction level.