Water Quality Classification and Spread Velocity Estimation using Deep Learning
Water pollution has always posed in a great threat to human health and aquatic life in the water bodies. Such pollution is certainly a major source of diseases like typhoid, diarrhea, jaundice and sometimes can also be fatal especially in the developing nations. So it is necessary to determine the probable contaminants that could be present in water. Therefore, this work focuses on the classification of water contamination based on color which would predict the probable contaminants so that it would be helpful to know what type of harmful contaminants could be present. For the classification of the contaminated water, a total of 850 images with only 6 colours viz. blue, black, brown, red, green and white were considered since the system had to classify into 6 different colour classes. This work mainly focused on the use of machine learning techniques so as to provide an efficient way to carry out the above stated problem. After succinct training with a dataset representing various colours of contaminated water , the deep learning algorithm (CNN) achieved an accuracy around 88%. Along with the water contamination, it was also equally important to address the impurities in water. So a unique method to calculate the spread velocity of liquid impurity in water using basic image processing technique was developed.