Abstract

Facial emotion recognition (FER) technology in autonomous vehicle drivers can considerably strengthen the efficiency and safety of the driving experience. The system can analyze facial expressions in real-time by employing advanced computer vision (CV) techniques, which identify emotions such as stress, fatigue, or distraction. This enables the vehicle to adapt its behavior, triggering interventions or alerts where applicable to alleviate possible threats. Ensuring the emotional well-being of the driver promotes a safer road environment, improving overall road safety and diminishing the possibility of accidents in the era of autonomous vehicles. FER using (Deep Learning) DL is an advanced technique that leverages deep neural network (DNN) to automatically interpret and identify emotions from facial expressions. DL algorithms, especially Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) have attained outstanding results in this field since they allow us to learn temporal dependencies hierarchy and features within the data. This research develops a novel Computer Vision with Optimal DL-based Emotion Recognition (CVODL-ER) model for Autonomous Vehicle Drivers. The CVODL-ER method concentrates on the automated classification of various sorts of emotions of autonomous vehicle drives. To accomplish this, the CVODL-ER technique makes use of the SE-ResNet model for learning intrinsic patterns from the driver's facial images. Besides, the hyper parameter tuning of the SE-ResNet model takes place via a quasi-oppositional Jaya (QO-Jaya) algorithm. For the recognition of driver emotions, the CVODL-ER system executes the deep belief network (DBN) algorithm. The performance analysis of the CVODL-ER technique takes place using a benchmark facial image database. The obtained results underline the improved efficiency of the CVODL-ER technique over other models.