Detekcija objekata modelima strojnog učenja

master's thesis

• Title (english): Object detection based on machine learning models
• Author: Filip Butić
• Mentor: Marko Čupić (mentor)
• Committee member: Marko Čupić (predsjednik povjerenstva)
• Committee member: Željka Mihajlović (član povjerenstva)
• Committee member: Tomislav Hrkać (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-02-22, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Računalni vid danas ima mnoge primjene kako u industriji tako i u znanosti. Detekcija objekata je samo jedna od područja računalnog vida te ima mnogo rješenja kao što su jednofazni i dvofazni detektori te detektori koji se ne temelje na dubokom učenju.

U okviru ovog rada, preuzet je skup podataka te je pretprocesiran i očišćen od nepotrebnih razreda kao što su pješaci. Nadalje, skup podataka je augmentiran da bi se proširio skup podataka i povećala robusnost modela. Potom je provedeno više pokušaja treniranja da bi se dobio model s najboljim Naposljetku su te izvučene registarske oznake poslane OCR modelu da bi se iščitala registarska oznaka.

Abstract (english)

Computer vision has many uses today in idustry as well as in science. Object detection is one of the use cases from that area and has many solutions like the one-stage and two-stage detectors as well as detector that are not deep learning models.

For the purpose of this thesis, a dataset has been downloaded, preprocessed and cleaned from other unnecessary labels such as pedestrians. Furthermore, the dataset has been augmented to increase its size and the robustness of the model. Multiple training sessions were made to get the optimal model. Finally, the extracted license plates would be sent to an OCR model so that the license plate can be read.

• Keywords: računalni vid; OCR; augmentacija podataka; duboko učenje; strojno učenje; CNN; R-CNN; SSD; HOG; Viola-Jones; detekcija objekata
• Keywords (english): computer vision; OCR; dana augmentation; deep learning; machine learning; CNN; R-CNN; SSD; HOG; Viola-Jones; object detection
• Language: croatian
• URN:NBN: urn:nbn:hr:168:144613
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-07-13 09:58:36