Real-Time Stress Detection via Facial Recognition Using VGG16 CNN: A Non-Intrusive Approach

Stress is slowly but surely becoming a huge problem in the modern world. The new generation of adults is reporting average stress levels higher than ever before. Stress causes conditions like obesity, asthma, heart disease, diabetes, and depression. These health outcomes can manifest as acute symptoms or chronic conditions as well in case of prolonged exposure to consistently high stress levels. The aim is to remedy this using Real Time Stress Detection. This Real-Time stress detection was implemented using a web app. Users are asked to turn on their camera and record their face for 5 seconds. This footage will be used to generate their stress Score ranging from 0-10. The Model VGG-16 is extremely apt for the application as an F1 score of 91% was achieved, along with a high recall value of 92%. Image augmentation also proved useful as it greatly boosted the accuracy of the model. Future work includes integrating physiological signals like heart rate to enhance accuracy and a mobile app for personalized stress-relief strategies.