Detekcija pješaka dubokim neuronskim mrežama

undergraduate thesis

• Title (english): Pedestrian Detection Using Deep Neural Networks
• Author: Lovre Antonio Budimir
• Mentor: Sven Lončarić (mentor)
• Committee member: Sven Lončarić (predsjednik povjerenstva)
• Committee member: Nikola Banić (član povjerenstva)
• Committee member: Ivana Zadro (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2018-07-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U posljednjih nekoliko godina napredni sustavi za pomoć u vožnji preuzimaju sve veću ulogu u sprječavanju prometnih nesreća uzrokovanih ljudskom nepažnjom. Jedna od glavnih funkcionalnosti takvih sustava je mogućnost detekcije pješaka. U sklopu ovog rada opisane su konvolucijske neuronske mreže i njihove glavne značajke. Proučene su poznate arhitekture konvolucijskih neuronskih mreža koje u posljednjih nekoliko godina postižu jako dobre rezultate u području klasifikacije i detekcije objekata na slikama. Na temelju proučenih arhitektura odabran je model koji je treniran i testiran na prikupljenoj bazi slika prometnih scena. Navedeni model zatim je implementiran u sustav za detekciju pješaka.

Abstract (english)

In recent years, advanced driving assistance systems have taken an increasingly significant role in preventing traffic accidents caused by human negligence. One of the main functionalities of such systems is the ability to detect pedestrians. This paper describes the convolutional neural networks and their key features. Known architectures of convolutional neural networks, which in recent years have achieved very good results in the field of classification and detection of objects in the images, have been researched. Based on the studied architecture, a model was selected, trained and tested on the collected image dataset of traffic scenarios. That model was used in an implementation of a pedestrian detection system.

• Keywords: detekcija pješaka; duboke neuronske mreže; konvolucijske neuronske mreže; klasifikacija; detekcija objekata; baza slika
• Keywords (english): pedestrian detection; deep neural networks; convolutional neural networks; classification; object detection; image dataset
• Language: croatian
• URN:NBN: urn:nbn:hr:168:165389
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-03-13 19:24:36