Water quality prediction using CNN

Abstract The interaction of solar radiation with the water level concentration and the elements of the water cause the water to have its characteristic hue. The alteration of the color of the water is reflective of the alteration of the water’s properties and the degree to which it is suitable for use. Due to disasters like floods, tsunami in the last few years and water pollution has been an increasing problem. In world the intake of contaminated water causes 40% of deaths. Drinking unclean water is not safe and in order to reduce the issue to a level of extent, prediction of water quality can be done before consuming. The process used in water plants is based on the parameters pH, turbidity, temperature, hardness etc., of water using filtration and the water quality prediction can also be done using IOT by including both hardware and software. This project mainly comprises the primary level of water prediction using machine learning. Based on the color and quality of water the system predicts whether the given water sample is suitable for drinking or any further use. Tensorflow, Keras and CNN are used to train the model to forecast the water quality prediction. This project is cost-effective and works efficiently and can be used as immediate and initial level of water quality check since image processing tool is used. This model of water quality prediction can be checked using mobile captured and Google earth images of water samples.