Celebrity face recognition in videos
Author: Feliks Gorišek
Mentor: dr. Zdravko Kačič univ. dipl. inž. el.
Comentor: asist. Danilo Zimšek univ. dipl. inž. tel.
Degree: 2.
Date: May, 2019
DKUM: FELIKS GORIŠEK
The master’s thesis belongs to the field of digital processing of digital images and video signals. Face detection is present today on almost all mobile high-end devices.
The master’s thesis belongs to the field of digital processing of digital images and video signals. Face detection is present today on almost all mobile high-end devices. The field of face recognition in video clips, even in commercial use, is not so well established, so we will explore this area. In the theoretical part of the master’s thesis we will obtain the theoretical basis for finding key points and finding matching faces. In the practical part of the master’s thesis, the defined theoretical face recognition system will be implemented as a service for recognizing celebrities in video recordings. We will make practical tests of usage, experimental results and analyze the usable value.
Open Source Computer Vision is an open source library that is free of charge for both academic and commercial use. It supports interfaces for C ++, Python and Java programming languages and can be installed on Windows, Linux, Mac, iOS and Android operating systems. It includes hundreds of image processing algorithms. It has a modular structure, which means that each module contains several libraries.
Facial expressions play an important role in human communication. Recognition of human emotions is usually done by recognizing facial expressions. Analyzing and recognizing the expression on the face is a complex task, as people express emotions differently, but at the same time the expression changes facial features, which further complicates and causes anomalies in face recognition.
Program flowchart
