Clasificación de Granos de Café usando FPGA

Abstract

This work presents the development of a system for the classification of coffee beans based on the maturity stages using Artificial Neural Network Systems (ANNs). As classification tool two RNAs\' structures were used, Multi-Layer Perceptron (MLP) and the block-based neural network model (BBNN). Multilayer perceptron structure (MLP) has been designed and implemented on C++ using the back-propagation learning algorithm. To increase the execution speed for ANN, it has been implemented on hardware using natural parallelism. The block-based neural network (BBNN) model consists of a two-dimensional array of fundamental blocks and integer weights in order to allow easier implementation with reconfigurable hardware such as field programmable gate array (FPGAs). The architecture is globally optimized using a genetic algorithm. This architecture has been implemented and synthesized on Altera Flex 10K FPGAs. The percentage of effectiveness for MLP structure was 91.7\\% and for BBNN model was 89.5\\%.