Stress Detection using Facial Expression

In this study, we present an advanced stress detection framework that automatically interprets facial expressions to assess stress levels. Addressing the limitations of earlier versions of this work—including vague methodology descriptions and a lack of quantitative evaluation— we now propose a hybrid system combining Convolutional Neural Networks (CNNs) for spatial feature extraction with Long Short-Term Memory (LSTM) units for temporal emotion tracking. Our method was rigorously evaluated using well-known datasets such as FER-2013, CK+, and AffectNet and achieved a mean classification accuracy of 92.15%. The system offers a non-invasive approach to stress assessment and significantly surpasses traditional machine learning benchmarks. Future developments will explore integrating physiological signals for a richer, multimodal stress evaluation system.