Čitanje tiskanog teksta pomoću dubokih neuronskih mreža

undergraduate thesis

• Title (english): Reading of printed text using deep neural networks
• Author: Luka Marković
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Ivana Zadro Matovinović (član povjerenstva)
• Committee member: Jelena Ivanković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-09-10, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom završnom radu opisano je kako funkcioniraju neuronske mreže i dan je uvid u njihovu arhitekturu. Konvolucijske neuronske mreže su detaljnije opisane jer su se koristile u izradi rada. Detaljno su opisani konvolucijski i potpuno povezani sloj. U radu je detaljno opisan sustav za prepoznavanje i klasifikaciju slika, YOLO (You only look once). Nakon teoretskog uvoda navedeno je kako koristiti sustav i pokazani su primjeri proizišli iz testiranja neuronske mreže. Slike prikazuju kako dobru klasifikaciju tako i greške i poteškoče na koje nailazi sustav.

Abstract (english)

This paper describes how neural networks work and gives insight into their architecture. Convolution Neural Networks are described in more detail because they were used for this paper. A convolutional and fully-connected layer is described in detail. This paper describes the system for recognizing and classifying images, YOLO (you only look once). After the theoretical introduction, it was stated how to use the system and demonstrated examples derived from the testing of neural network. The pictures show how good the classification is but also the mistakes and difficulties faced by the system.

• Keywords: duboka neuronska mreža; konvolucijska neuronska mreža; YOLO; OCR; konvolucija; TensorFlow; DarkFlow
• Keywords (english): deep neural networ; convolutional neural network; YOLO; OCR; convolution; TensorFlow; DarkFlow
• Language: croatian
• URN:NBN: urn:nbn:hr:168:700297
• 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: 2020-01-07 20:04:58