prikaz prve stranice dokumenta Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma
  Download
PDF 5.7 MB
master's thesis
Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma
2022. urn:nbn:hr:190:358489
Cvija, Tajana
University of Rijeka
Faculty of Engineering
Department of Automation and Electronics

Cite this document

Cvija, T. (2022). Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma (Master's thesis). Rijeka: University of Rijeka, Faculty of Engineering. Retrieved from https://urn.nsk.hr/urn:nbn:hr:190:358489

Cvija, Tajana. "Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma." Master's thesis, University of Rijeka, Faculty of Engineering, 2022. https://urn.nsk.hr/urn:nbn:hr:190:358489

Cvija, Tajana. "Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma." Master's thesis, University of Rijeka, Faculty of Engineering, 2022. https://urn.nsk.hr/urn:nbn:hr:190:358489

Cvija, T. (2022). 'Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma', Master's thesis, University of Rijeka, Faculty of Engineering, accessed 05 May 2024, https://urn.nsk.hr/urn:nbn:hr:190:358489

Cvija T. Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma [Master's thesis]. Rijeka: University of Rijeka, Faculty of Engineering; 2022 [cited 2024 May 05] Available at: https://urn.nsk.hr/urn:nbn:hr:190:358489

T. Cvija, "Detekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma", Master's thesis, University of Rijeka, Faculty of Engineering, Rijeka, 2022. Available at: https://urn.nsk.hr/urn:nbn:hr:190:358489

Please login to the repository to save this object to your list.
Metadata
TitleDetekcija karcinoma mokraćnog mjehura korištenjem YOLO algoritma
Title (english)Urinary bladder cancer detection using YOLO algorithm
AuthorTajana Cvija
MentorZlatan Car (mentor)
Committee memberSaša Vlahinić (predsjednik povjerenstva)
Committee memberZlatan Car (član povjerenstva)
Committee memberIvan Volarić (član povjerenstva)
GranterUniversity of Rijeka
Faculty of Engineering
(Department of Automation and Electronics)
Rijeka
Defense date and country2022-09-27, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Electrical Engineering
Automation and Robotics
Abstract
Ovaj diplomski rad opisuje detekciju karcinoma mokraćnog mjehura primjenom YOLO (You Only Look Once) algoritma. Opisana je problematika računalnog vida i detekcije objekata te uloga konvolucijskih neuronskih mreža pri rješavanju iste. Konvolucijske neuronske mreže opisane su u pogledu strukture, procesa učenja i vrednovanja detekcije objekata. Predstavljena su rješenja detekcije objekata temeljena na konvolucijskim neuronskim mrežama u obliku algoritama s pristupom u dva i u jednom koraku. Iz algoritama s pristupom u jednom koraku izdvojen je i pobliže opisan YOLO algoritam s naglaskom na svoju najnoviju inačicu YOLOv7. YOLOv7 model učen je detekciji tumora na mokraćnom mjehuru na skupovima podataka koji su se sastojali od CT i MRI slikovnih zapisa frontalnog, horizontalnog i sagitalnog presjeka trbušne šupljine. Učenje za pojedini presjek vršeno je odvojeno i rezultiralo je srednjom prosječnom preciznošću od 94.4%, 85.6% i 96.1% za frontalni, horizontalni i sagitalni presjek, redom. Ispitivanjem modela na novim, modelu nepoznatim, slikama utvrđena je uspješna detekcija karcinoma mokraćnog mjehura.
Abstract (english)
This master's thesis describes urinary bladder cancer detection using YOLO (You Only Look Once) algorithm. It describes the problem of computer vision, object detection, and the role of using convolutional neural networks as its solution. Convolutional neural networks are described in the means of their structure, training process, and object detection evaluation. Solutions of object detection based on convolutional neural networks are introduced as one- and two-stage approach algorithms. From one-stage approach algorithms YOLO algorithm has been chosen and described in more detail with emphasis on its newest version YOLOv7. A YOLOv7 model has been trained to detect tumor on the urinary bladder by using datasets that consisted of CT and MRI images of frontal, horizontal and sagittal plane of the abdomen. Learning process for each plane has been performed separately and it has resulted in 94.4%, 85.6%, and 96.1% mean average precision for frontal, horizontal and sagittal plane, respectively. Model testing on new, to the model unknown, images concluded in successful detection of the urinary bladder cancer.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:190:358489
Study programmeTitle: Graduate University Study of Electrical Engineering
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike (magistar/magistra inženjer/inženjerka elektrotehnike)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence Attribution 4.0 International (CC BY 4.0) icon
Created on2022-09-20 18:50:34