Machine learning-based visual health monitoring system for garment factory workers

Abstract

Garment factory workers are at high risk for health hazards due to their regular exposure to synthetic dyes and harmful chemicals, which can lead to various skin diseases. This study introduces a machine learning-based visual health monitoring system to address these challenges. The system employs advanced image processing techniques to analyze images of workers’ skin, allowing for the detection and evaluation of the stage and severity of any condition. In addition to image analysis, a short survey will be conducted with garment workers to collect information about their symptoms, pain levels, and how severe they perceive their skin problems. The image analysis and the survey responses will then be combined into a robust machine learning framework. To improve the reliability of diagnosis, the research utilizes convolutional neural networks (CNNs) and Transformer models, which help address potential variations in image quality. For training the models, a diverse dataset of skin disease images has been gathered from sources such as HAM10000, Kaggle, and various other online platforms. These data sets are a solid foundation for creating reliable models for accurate detection. In addition, the research includes creating a mobile application designed to support real-time diagnostics and monitoring. This study aims to improve early skin disease detection and establish an effective health monitoring system for garment factory workers, ultimately contributing to their overall health and well-being.