A Connectivity Aware Graph Neural Network for Real Time Drowsiness Classification

Drowsiness detection performs a important role in improving driving force protection and preventing accidents because of fatigue. Our technique integrates the blessings of several superior gadget mastering algorithms to enhance prediction accuracy and responsiveness. Specifically, we rent a Graph Neural Network to model the spatial and temporal dependencies in driving force conduct, coupled with a Recurrent Neural Network architecture the usage of Gated Recurrent Units to capture long-time period sequential styles. Furthermore, the XGBoost algorithm is applied for function enhancement, and Random Forest is used to offer an ensemble learning framework for robust class. The CAGNN framework is designed to dynamically alter to real-time changes in connectivity and automobile surroundings, ensuring seamless performance even in varying conditions. Experimental results show that our version appreciably outperforms conventional drowsiness detection methods in phrases of accuracy, latency, and adaptableness to actual-world situations.