Water Contamination Detection Using Image Processing

Water contamination, and pollution in the water bodies has affected the human life and danger to the aquatic life. Many Countries have the strict regulations to prevent the water contamination. It is important to monitor the water bodies for the sustainable life around the human beings and animals. This chapter focuses on the use of recent advancements in image processing and machine learning techniques to detect water pollution. The color-based water pollution detection model is developed to predict the possible pollutants in the water. The water pollution ratio of the surface area is calculated using canny edge detectors and contour area marking techniques. The surface area covered by plastic bottles is marked and displayed as a percentage of pollution. Three different ML Models are used on the non-image dataset to predict the water quality. The non-deep learning model has shown accuracy of 97.53%. The deep learning based linear regression model has shown the loss of 0.0130 and Mean Square Error of 0.0130 after 50 epochs. The logistic regression has achieved an accuracy of 94.76%.