In this work we present a computer vision based colorimetric method for measuring concentration of potassium permanganate in water. Images of liquid samples are captured by a micro camera and processed by Raspberry Pi microcomputer. A customized computer vision application developed in python extracts gray scale value of the images. Non-linear least square regression analysis reveals that gray scale value decreases exponentially with increasing concentration of potassium permanganate. Based on the regression model, the proposed method is capable to measure concentration of potassium permanganate very precisely with resolution ranging from 0.62 mg/L – 6.2 mg/L when concentration is varied from 0 to 4.0 mM. This result has confirmed that the proposed method is good enough to detect do not consume limit of potassium permanganate, which is 7.0 mg/L. Performance of the system is validated by using 40 liquid samples. Novelty of the proposed method includes low-cost, portability, real time capability and scalability to work with other colored chemical solutions.