Multi-State Emotion and Drowsiness Detection Using Hybrid EEG Deep Learning Models
This project focuses on developing an advanced system that detects human emotions and drowsiness levels using electroencephalogram (EEG) signals. A hybrid deep learning architecture combining Convolutional Neural Networks (CNNs), Bidirectional Long Short-Term Memory (BiLSTM) networks, and Self-Attention mechanisms was implemented to effectively capture spatial, temporal, and contextual features of brainwave activity.
The system was trained on benchmark EEG datasets and achieved high accuracy in classifying multiple emotional states (happy, sad, fearful, neutral) as well as drowsiness levels (alert, moderate, sleep onset). The model integrates robust preprocessing techniques such as artifact removal, noise reduction, and normalization, along with metaheuristic optimization for fine-tuning.
Multi-State Detection: Simultaneously recognizes emotions and drowsiness in real time.
Hybrid Architecture: CNN for spatial feature extraction, BiLSTM for temporal dynamics, and Attention for interpretability.
High Performance: Achieved ~82% accuracy in both emotion recognition and drowsiness detection.
Applications: Driver fatigue monitoring, mental health assessment, human-computer interaction, and smart learning environments.
This work contributes to the fields of affective computing and brain-computer interfaces (BCIs), paving the way for real-world applications that enhance safety, well-being, and adaptive human-computer collaboration.
At Lendi, we are recognized as a top institution dedicated to excellence in education.we offer exceptional academic programs that prepare our students for global success. Discover the unique qualities that make our students stand out.
©Lendi Institute of Engineering and Technology - Established 2008