FPGA implementation of a Convolutional Neural Network for image classification

In this paper, a technique for implementing a Convolutional Neural Network (CNN) is presented. The implementation is performed on a reconfigurable Field Programmable Gate Array (FPGA) circuit of the PYNQ-Z1 platform. The objective is to improve the processing performance for embedded applications. The Theano Lasagne framework and the BNN-PYNQ library have been used to develop the CNN model, called Fruits_Classifier, using Python programming language. The performances of this technique were evaluated in terms of accuracy and cost. Simulation results showed the efficiency of this technique and validate the proposed design approach. The proposed technique is ready for a hardware FPGA implementation and it can be very useful for real-time embedded applications.