Mobilna aplikacija za prepoznavanje poveznica na slikama korištenjem dubokog učenja

master's thesis

• Title (english): Mobile Application for Recognizing Links in Images Using Deep Learning
• Author: Andrej Ceraj
• Mentor: Vladimir Čeperić (mentor)
• Committee member: Vladimir Čeperić (predsjednik povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Committee member: Tomislav Marković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-15, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Prepoznavanje teksta sa slike poznat je problem u računalnom vidu. Dosad postoji nekoliko pristupa za rješavanje tog problema, no u ovom radu proučava se rješenje korištenjem dubokog učenja. Modeli koji su se koristili prilikom prepoznavanja teksta bili su Yolo za detektiranje teksta i CRNN za prepoznavanje teksta iz detekcije. Nad ovim modelima, osim točnosti, stavljen je zahtjev da se moraju moći koristiti na mobilnom uređaju i da se rezultat može dobiti u skoro stvarnom vremenu. Opisana je arhitektura modela, podaci za treniranje, metode treniranja, rezultati i potencijalna poboljšanja. Prikazano je korištenje tih modela u mobilnoj aplikaciji za prepoznavanje poveznica te rad same aplikacije.

Abstract (english)

Recognizing text from an image is a common problem in computer vision. So far, there have been several approaches to solving this problem, but this paper examines the solution using deep learning. Models used in solving this problem were Yolo for text detection and CRNN for text recognition from the detection. The models, in addition to accuracy, are required to be able to run on mobile devices and in near real time. Architecture, training data, training methods, results and potential improvements are described bor both models. Finally, this paper shows the use of these models in the mobile application for link recognition and the operation of the application itself.

• Keywords: duboko učenje; računalni vid; detekcija objekata; prepoznavanje teksta; Yolo; CRNN; mobilna aplikacija
• Keywords (english): deep learning; computer vision; object detection; text recognition; Yolo; CRNN; mobile application
• Language: croatian
• URN:NBN: urn:nbn:hr:168:492190
• 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 11:57:28