Primjena FPGA programabilne logike za ubrzanje rada neuronske mreže za prepoznavanje gesti ruke

master's thesis

• Title (english): Application of FPGA programmable logic to accelerate neural network operation for hand gesture recognition
• Author: Nikola Jerković
• Mentor: Zdenko Kovačić (mentor)
• Committee member: Zdenko Kovačić (predsjednik povjerenstva)
• Committee member: Stjepan Bogdan (član povjerenstva)
• Committee member: Matko Orsag (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-02-19, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U radu su obrađene mogućnosti implementacije duboke neuronske mreže na Xilinx edukacijskoj platformi PYNQ Z1. Prolazi se kroz načine rada različitih rješenja, njihove prednosti, mane, načine postavljanja na platformu i poteškoće koje dolaze s tim. Krajnji rezultati su dijelom dobiveni eksperimentom, dijelom teoretski te se uspoređuju s alternativnom platformom nVidia Jetson TX2 koja umjesto FPGA sklopa koristi nVidia GPU s CUDA akceleracijom. Jetson TX2 se za svoju cijenu pokazao drastično jačim, konzistentniji u performansama te daleko lakši za razvoj i eksperimentaciju s puno širom podrškom.

Abstract (english)

In this work we look at the possible implementations of deep neural networks on Xilinx's educational platform, the PYNQ Z1. We go through the way different solutions work, their advantages and disadvantages, setup procedures and the difficulties that come with them. The end results are part eksperiment, part theory and are compared to an alterntative platform, the nVidia Jetson TX2 which uses an nVidia GPU with CUDA acceleration instead of an FPGA circuit. The Jetson TX2 proved to be drastically more powerful, consistent in performance and far easier to develop and experiment with while alos having far better support.

• Keywords: FPGA; duboke neuronske mreže; Vitis AI; FINN; Jetson TX2; DPU
• Keywords (english): FPGA; deep neural networks; Vitis AI; FINN; Jetson TX2; DPU
• Language: croatian
• URN:NBN: urn:nbn:hr:168:206925
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 10:06:54