prikaz prve stranice dokumenta Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika
Access restricted to authenticated users
master's thesis
Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika
Split: University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, 2021. urn:nbn:hr:179:154092
Krstanović, Delfa
University of Split
Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture
Department of Electronics and Computing

Institutional repository: Repository of the Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture

Cite this document

Krstanović, D. (2021). Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika (Master's thesis). Split: University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture. Retrieved from https://urn.nsk.hr/urn:nbn:hr:179:154092

Krstanović, Delfa. "Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika." Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, 2021. https://urn.nsk.hr/urn:nbn:hr:179:154092

Krstanović, Delfa. "Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika." Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, 2021. https://urn.nsk.hr/urn:nbn:hr:179:154092

Krstanović, D. (2021). 'Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika', Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, accessed 16 January 2025, https://urn.nsk.hr/urn:nbn:hr:179:154092

Krstanović D. Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika [Master's thesis]. Split: University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture; 2021 [cited 2025 January 16] Available at: https://urn.nsk.hr/urn:nbn:hr:179:154092

D. Krstanović, "Primjena metoda umjetne inteligencije za klasifikaciju medicinskih slika", Master's thesis, University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, Split, 2021. Available at: https://urn.nsk.hr/urn:nbn:hr:179:154092

Metadata
TitlePrimjena metoda umjetne inteligencije za klasifikaciju medicinskih slika
Title (english)Application of artificial intelligence methods for classification of medical images
AuthorDelfa Krstanović
MentorJosip Musić https://orcid.org/0000-0002-9185-5762 (mentor)
GranterUniversity of Split
Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture
(Department of Electronics and Computing)
Split
Defense date and country2021-09-21, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Electrical Engineering
Abstract
Pojam klasifikacije slike odnosi se na analizu i identifikaciju klase kojoj ta slika pripada. Za klasifikaciju slika uglavnom se koriste konvolucijske neuronske mreže. U radu su klasificirane medicinske slike oboljenja kože, koje su podijeljene u dvije klase: maligno i benigno. Za klasifikaciju su korištene tri standardne arhitekture neuronskih mreža: MobileNet, ResNet50 i InceptionV3. Sve su mreže trenirane prvo na izvornom skupu podataka. Zbog neravnoteže među klasama, klasa benigno ima puno više slika od klase maligno, rezultati tijekom treninga nad takvim podacima su lošiji u pogledu klasifikacije klase maligno. Nakon provedene augmentacije MobileNet i ResNet50 pokazuje prihvatljive rezultate. Za razliku od njih model InceptionV3 postiže u oba slučaja loše rezultate.
Abstract (english)
The term image classification refers to the analysis and identification of the class to which that image belongs. Convolutional neural networks are mainly used for image classification. The paper classifies medical images of skin diseases, which are divided into two classes: malignant and benign. Three standard neural network architectures were used for classification: MobileNet, ResNet50, and InceptionV3. All networks were trained first on the original data set. Due to the imbalance between classes, the benign class has a lot more images than the malignant class, the results during training over such data are worse in terms of malignant class classification. After augmentation, MobileNet and ResNet50 show acceptable results. In contrast, the InceptionV3 achieves poor results in both cases.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:179:154092
Study programmeTitle: Systems and Control
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka automatike i sustava (magistar/magistra inženjer/inženjerka automatike i sustava)
Type of resourceText
File originBorn digital
Access conditionsAccess restricted to authenticated users
Terms of useLicence In copyright icon
Created on2022-06-14 11:31:19