DESIGN AND IMPLEMENTATION OF TWO-DIMENSIONAL CONVOLUTION ON PYNQ-Z2 FPGA DEVELOPMENT BOARD

Abstract

Two-dimensional (2D) convolution is a very important operation commonly used in the fields of image processing and convolution neural networks. In this paper, we designed and implemented a hardware module that performs two-dimensional convolution for use in high-speed image processing. The convolution module was developed using hardware description language VHDL, synthesized on Xilinx's PYNQ-Z2 development board, and packed into a hardware library for use in Python development environments for related applications. Evaluation results showed that using the designed two-dimensional convolution module could improve the performance of the convolution operation by a factor of up to 9 times compared with the performance of the software implementation. The design has shown its potential in implementing FPGA-based hardware designs for image processing, pattern recognition, and deep learning applications.