Prijenosno učenje neuronske mreže ResNet za prepoznavanje pasmine pasa

undergraduate thesis

• Title (english): Transfer Learning of a ResNet Neural Network to Recognize Dog Breed
• Author: Ana Kikerec
• Mentor: Tomislav Petković (mentor)
• Committee member: Tomislav Petković (predsjednik povjerenstva)
• Committee member: Matea Đonlić (član povjerenstva)
• Committee member: Kristijan Bartol (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Cilj ovog rada je korištenje ResNet (Residual Network) neuronske mreže u svrhu raspoznavanja pasmine pasa. Koristi se baza slika Stanford Dogs Dataset, koja se sastoji od oko 20 000 slika pasa, podijeljenih u 120 vrsta. U radu najprije su opisane bitne značajke konvolucijskih mreža s fokusom na ResNet50 te su opisani alati korišteni za treniranje mreže. Praktični dio rada uključuje pripremu baze slika pasa, programsku implementaciju mreže, te učenje i testiranje mreže. U konačnici, evaluirani su rezultati učenja i uspješnost mreže.

Abstract (english)

The aim of this thesis is to use ResNet (Residual Network) neural network for the purpose of classifying dog breeds. The Stanford Dogs Dataset, which consists of around 20,000 dog images, is divided into 120 breeds. The essential features of convolutional neural networks are described, with the focus on ResNet50 and the tools used to train the network. The practical part of the work involves the preparation of a dog image database, the network deployment and ultimately the learning and testing of the network. Finally, learning outcomes and the effectiveness of the network were evaluated.

• Keywords: prijenosno učenje; neuronske mreže; konvolucijske neuronske mreže; konvolucija; ResNet; klasifikacija slika
• Keywords (english): transfer learning; neural networks; convolutional neural networks; convolution; ResNet; image classification
• Language: croatian
• URN:NBN: urn:nbn:hr:168:059400
• 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 19:05:07