USING MACHINE LEARNING AND IMAGE RECOGNITION TO EVALUATE WATER QUALITY

Abstract

This research project explores the application of machine learning in assessing water quality through image 
recognition. The study leverages a diverse dataset of water samples collected from various sources across Mumbai, encompassing 
ponds, lakes, water outlets, and sewage points. Multiple parameters, including pH, conductivity, turbidity, and dissolved oxygen 
content, are examined in relation to the resulting HEX colour code,which serves as a visual representation of water quality.The 
central hypothesis posits that employing machine learning algorithms can reliably predict water safety based on these 
environmental parameters through image recognition of the colour gradient of the water samples. A systematic approach to data 
collection, standardisation, and logistic regression modeling has been employed.Results demonstrate the effectiveness of the 
logistic regression model in predicting water safety with an 85% accuracy rate, highlighting its potential for real-time water quality 
monitoring and risk assessment. Nevertheless, this study recognises its limitations and the need for further research to refine the 
model's predictive accuracy and address variations across different geographical regions and water sources. This research aims to 
contribute to the development of innovative approaches for water quality assessment and therefore, environmental preservation.