The existence of hazardous chemicals and pollutants in water is one of many societal concerns. This water when consumed or used for some specific domestic or industrial purposes brings about apprehension regarding health-related consequences. This work explores the capability of a machine learning-based system in detecting the presence of heavy metals like Mercury (Hg) in water. It demonstrates near-accurate concentration prediction in the range of 0.001mg/liter to 100mg/liter using a voting regressor. Implementation of the system is done by integrating a camera with Raspberrypi, programmed to detect the presence of metals with its concentration and usability. Data set was prepared for training our model considering the different concentrations of reagents and water samples. This system demonstrates a good capability to predict the concentration of heavy metals along with its usage recommendations.

Colorimetry, Heavy Metal detection, Raspberry-Pi, Machine learning, Voting Regressor