Abstract:

Stress is a common problem today, and it can lead to various health issues. Early stress identification enables people to regulate their stress levels and stop the emergence of associated health issues. Recent years have seen a rise in the application of image processing and machine learning approaches to identify stress from a variety of physiological signals, including facial expressions, speech patterns, and heart rate variability. In this study, we suggest an image processing and machine learning-based stress detection system. The system takes pictures of people's faces and uses image processing methods to extract important details like skin tone, eye movement, and facial expressions. After that, a machine learning model, like a convolutional neural network, is trained using these features to classify image data into stressed and non-stressed categories. To develop and test the proposed system, we collected a dataset of facial images from individuals exposed to different levels of stress. The images were collected using a camera in a controlled environment. We then pre-processed the images to remove noise and extract relevant features using open-source software libraries such as OpenCV. We used these features to train a machine learning model and evaluate its performance using various metrics such as accuracy, precision, recall, and F1-score.

Description:

Convolutional neural network, Stress detection, Stress, Image processing, Machine Learning, Feature extraction.