This article describes the implementation of a computer vision system to determine the effect of time and temperature on the color generated by the roasting of coffee, for which a software tool in the mathematical software Matlab, previously parameterized using color data for the roasting process was developed and implemented. The color was measured in CIEL*a*b* space and the data were analyzed using the statistical response surface design. The result of this work is a system that allows for realtime sensory information about color and therefore has an advantage over existing traditional systems. The study shows that computer vision system could distinguish different shades during roasting of coffee depending on process parameters and good capacity for generalization.