Duboki konvolucijski modeli za lokalizaciju objekata

master's thesis

• Title (english): Deep convolutional models for object localization
• Author: Damjan Miko
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Committee member: Slobodan Ribarić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2018-07-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad bavi se metodama detekcije objekata konvolucijskim modelima. Detaljnije su proučeni i uspoređeni modeli SSD i YOLO te njihove novije verzije. Oni se baziraju na jednom unaprijednom prolazu kroz mrežu što ih čini brzim, a također daju i dobre rezultate. Napravljen je eksperiment koji koristi novije YOLO modele za lokalizaciju objekata na vlastitom skupu podataka. U tu svrhu koristi se snimka nogometne utakmice gdje su objekti za detekciju igrači oba tima i sudac. Modeli su evaluirani... More osnovnim mjerama i prikazani su njihovi rezultati detekcije. Less

Abstract (english)

This paper deals with the object detection method with convolutional models. There are studied and compared models like SSD, YOLO and their modern versions. They are based on only one forward pass through network what makes them fast and also give good results. An experiment is made which use newer YOLO models for localization objects on own dataset. For this purpose a tape of football game is use where object for detection are both teams and referee. Models are evaluated with basic measures... More and their detection results are shown. Less

• Keywords: detekcija objekata; konvolucijske neuronske mreže; duboko učenje; SSD; DSSD; YOLO
• Keywords (english): object detection; convolutional neural networks; deep learning; SSD; DSSD; YOLO
• Language: croatian
• URN:NBN: urn:nbn:hr:168:135574
• 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: 2019-03-13 23:05:18