Primjena modela neuronske mreže na FSOCO skupu slika

undergraduate thesis

• Title (english): Application of neural network models to the FSOCO image set
• Author: Nikolina Sitar
• Mentor: Tomislav Stipančić (mentor)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2022-09-28, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Mechanical Engineering

Abstract

U ovom radu dan je teorijski pregled strojnog učenja, konvolucijske neuronske mreže te pojmova bitnih za razumijevanje same evaluacije dobivenih rezultata. Nakon prikladnog uvođenja pojmova, rad se nastavlja s praktičnim dijelom zadatka koji kreće s odabirom prikladnog modela postojeće mreže. Nastavno na to, odabrani se model trenira na FSOCO skupu podataka. Navedeni skup podataka je jedini prikladan za ovaj zadatak zbog specifičnosti objekata koje je potrebno detektirati i klasificirati. Na kraju se provodi evaluacija dobivenih rezultata u kontekstu buduće primjene treniranog modela, s fokusom na vrijednosti srednje prosječne preciznosti (mAP).

Abstract (english)

In this paper, theoretical background is given regarding the machine learning, convolutional neural networks and main expressions crucial for understanding the evaluation of the results. After appropriatelly introducing key expressions, paper is continued with practical part which starts with choosing the right model of existing network. Afterwards, chosen model is being trained on FSOCO dataset. Mentioned dataset is the only appropriate for this problem due to specific object that need to be detected and classified. At the end, evaluation of the results is made keeping in mind future use of trained model, with focuse on mean average precision (mAP).

• Keywords: strojno učenje; konvolucijska neuronska mreža; FSOCO; detekcija; klasifikacija; srednja prosječna preciznost
• Keywords (english): machine learning; convolutional neural networks; FSOCO; detection; classification; mean average precision
• Language: croatian
• URN:NBN: urn:nbn:hr:235:663552
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka strojarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Embargoed access; Embargo expiration date: 2024-09-22
• Created on: 2022-09-23 10:58:29