ColorPro

← Back to Portfolio

Roasting coffee transforms the chemical and physical properties of green coffee beans making then roasted coffee products. The roasting process is what produces the characteristic flavour of coffee by causing the green coffee beans to expand and to change in colour, taste, smell, and density. The quality of the produced coffee’s characteristics is extremely dependent on the degree of roast, which is determined by the bean’s colour, itself identified by the eye of an experienced specialist.

ColorPro was created to help in this process, the main objective being to automate the process and to eliminate the dependency of a specialist. ColorPro is a system that improves the efficiency and quality of the coffee roasting process. It receives as an input sample images of the coffee beans and parameters of grinding and roasting machines, such as temperature, and roast type, among others.

After the system is well configured, the user can capture an image at any stage of the roasting process. With that image, ColorPro selects important areas from which to analyse features. Thus, the user is informed as to the remaining time for which the roasting process should occur. From this approach, the produced coffee will always achieve a high consistent quality, without the requirement of an experienced specialist.

This software was developed for SENAI-RJ / FIRJAN, and was very well received by clients and brought clear value. This is illustrated by the participation of ColorPro in one of the largest trade events for the food and beverage industry in Latin America: "TecnoBebida".

This personal project was undertaken with a colleague with its design and development from scratch in 2009. Principally based on colorimetry and neural networks techniques, other technologies involved were: C++, Iup and Lua.


Project Details
  • Project name: ColorPro
  • Category: Professional Projects
  • Period: Jan 2009 - Nov 2009
  • Role: Contractor Software Engineer
  • Worked for: SENAI-RJ and FIRJAN
Technologies
C++ Computer Vision Machine Learning Lua