Sustav za raspoznavanje prometnih znakova

undergraduate thesis

• Title (english): System for Traffic Sign Recognition
• Author: Matej Dobrovodski
• Mentor: Zoran Kalafatić (mentor)
• Committee member: Zoran Kalafatić (predsjednik povjerenstva)
• Committee member: Darijan Marčetić (član povjerenstva)
• Committee member: Martin Soldić (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 radu su opisani popularni pristupi za klasifikaciju objekata s naglaskom na korištenje konvolucijskih neuronskih mreža. Objašnjena je generalna ideja konvolucijskih neuronskih mreža te je opisan postupak izgradnje jedne jednostavne mreže u svrhu raspoznavanja prometnih znakova korištenjem Python biblioteke Keras. Izgrađeni sustav primijenjen je na javno dostupnom skupu slika prometnih znakova GTSRB s relativno visokom točnošću od 95% i više, bez potrage za optimalnim hiperparametrima. Dan je osvrt na cjelokupni sustav, rezultate te opis smjera u kojem bi se razvoj sustava nastavio u svrhu još veće točnosti i primjene u stvarnom svijetu.

Abstract (english)

This paper describes well-known approaches to classification of objects with emphasis on the use of convolutional neural networks. The general idea behind convolutional neural networks is explained and a procedure is given for constructing a single simple network for the purpose of recognizing traffic signs using the Python library Keras. The built system was applied to the publicly available GTSRB traffic sign dataset with a relatively high accuracy of 95% and above, without a search for optimal hyperparameters. An overview of the entire system is given, as well as the results and the description of the direction in which the development of the system would continue for the purpose of even greater accuracy and application in the real world.

• Keywords: prometni znak; klasifikacija; konvolucijske neuronske mreže; sustav za raspoznavanje; računalni vid; GTSRB; Keras
• Keywords (english): traffic sign; classification; convolutional neural networks; system for recognition; computer vision; GTSRB; Keras
• Language: croatian
• URN:NBN: urn:nbn:hr:168:307322
• 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 19:56:56