Detekcija i označavanje zuba u panoramskim rendgenskim snimkama pomoću Faster-RCNN arhitekture

undergraduate thesis

• Title (english): Detection and Classification of Teeth in Panoramic X-ray Images Using Faster-RCNN Architecture
• Author: Karlo Koledić
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Ivana Zadro (član povjerenstva)
• Committee member: Jelena Ivanković (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 ovom radu je opisan problem detekcije i klasifikacije zuba na rendgenskim slikama. Navedene su metode, modeli i arhitekture korištene pri rješavanju problema s posebnim osvrtom na Faster R-CNN arhitekturu.

Na početku su objašnjene metode dubokog učenja u računalnom vidu poput gradijentnog spusta, propagacije pogreške unatrag te konvolucijskih neuronskih mreža. Zatim su opisane metode i modeli koji se koriste za rješavanje problema detekcije objekata s naglaskom na Faster R-CNN. Objašnjena je struktura te poboljšanja Faster R-CNN-a s obzirom na prethodne arhitekture.

Konačno je Faster R-CNN arhitektura primijenjena na problem detekcije i klasifikacije zuba na rendgenskim slikama. Nakon treniranja mreže, navedeni su i objašnjeni rezultati te mogući uzroci razlike u preciznosti modela s obzirom na različite parametre.

Abstract (english)

This paper describes the problem of detection and classification of teeth on x-ray images. The methods, models, and architectures used in solving problems with a special focus on the Faster R-CNN architecture are listed.

At the beginning, explanation of deep learning methods in computer vision is provided, such as gradient descent, backpropagation algorithm, and convolutional neural networks. Then, the methods and architecture that are used to solve the object detection problem with the emphasis on the Faster R-CNN architecture are described. Architecture and improvements to Faster R-CNN have been explained with regard to previous architectures.

Finally, the Faster R-CNN architecture was applied to the problem of detecting and classifying teeth on X-ray images. After the training of the network, the results and possible causes of the differences in model precision are listed with regard to different parameters.

• Keywords: računalni vid; duboko učenje; konvolucijske neuronske mreže; detekcija objekata; Faster R-CNN
• Keywords (english): computer vision; deep learning; convolutional neural networks; object detection; Faster R-CNN
• Language: croatian
• URN:NBN: urn:nbn:hr:168:142886
• 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 21:39:28