Advancing Water Safety: Exploring AI for Escherichia coli Detection in Drinking Water

Ensuring the availability of safe drinking water is essential for the overall health of the public, especially in the rural regions of West Bengal, where dependence on wells, rivers, and ponds raises concerns about potential water contamination. The problem is worsened by the absence of inexpensive testing equipment, given that traditional laboratory testing is a time-consuming process. To address this, we propose a portable, cost-effective water testing system utilizing image processing and artificial intelligence (AI) to swiftly detect E. coli presence, a key indicator of water safety. Our unique approach involves employing image-processing algorithms to generate input data for machine-learning models, with VGG16 identified as the most effective, achieving 99% accuracy through rigorous experimentation with diverse water samples. The pivotal contribution includes a substantial dataset of 1000 E. coli images from 200 water samples, supplemented by 300 online-sourced images. Our image preparation techniques, including gray scaling, noise reduction, and contour detection, significantly improve dataset quality and enhance machine-learning model effectiveness. This research, encompassing dataset creation, image preprocessing, and model selection, provides a comprehensive exploration of E. coli detection in water samples. We anticipate that our methodologies will advance AI-based E. coli detection systems, offering broader implications for environmental and public health monitoring.