FPGA implementation of an images classifier

Artificial intelligence is one of the main topic in nowadays scientific researches and it is starting to be appealing also for industrial productions. Due to the increase of interest of the industries, Xilinx, one of the major hardware supply companies, decided to create a development flow that allows to accelerate the production of artificial intelligence based systems. This thesis is centered on the study of the tool that allows to use the Xilinx new development flow, called Vitis AI. An overview on its functioning and its main component will be provided together with a project example whose main purpose is exploring the practical usage of the tool-chain. In addition to that, the project has been extended to include the design of a hardware accelerator that allows to increase the performances of the developed application by transforming a software function into a hardware component.