Detection of Drivers behavior using Real time drowsiness classification in Machine Learning Methods

Authors- Mr J Jagadeswara Reddy, Swathi Devarangula, Harsha Priya, Veera Raja Mohan Reddy, Birru Madhu

Abstract-– Real-time drowsiness classification, aiming to improve the detection of drowsiness in environments where safety is critical, such as in driving or machine operation. The model processes various physiological and behavioral signals, including eye movements, facial expressions, and heart rate, by representing them as nodes in a graph. The edges between nodes capture the inter dependencies among these signals, allowing the model to better understand the complex patterns associated with drowsiness. By incorporating connectivity-awareness, the CAGNN enhances the detection process by prioritizing key sensor interactions and adapting to changing input conditions. This approach enables more accurate classification, even in the presence of noisy or incomplete sensor data, which is a challenge for traditional methods. By leveraging the relationships between multiple signals, the CAGNN is able to provide more robust and dynamic real-time drowsiness assessments. Experimental results show that the CAGNN outperforms traditional methods in both accuracy and computational efficiency. The model is designed to deliver low-latency predictions, ensuring fast and reliable real-time performance. This approach provides an effective solution for drowsiness detection, improving safety and reducing risks associated with drowsiness in various operational settings.

DOI: /10.61463/ijset.vol.13.issue2.288