Očitavanje pulsa osobe iz videosnimke

master's thesis

• Title (english): Estimating the Person's Pulse From a Video Recording
• Author: Ena Rajković
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Sven Lončarić (član povjerenstva)
• Committee member: Krešimir Križanović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Cilj ovog rada je detektirati puls iz videosnimke u boji. Nakon prikupljanja snimki, predstavljeno je rješenje uz korištenje duboke neuronske mreže za detekciju promjena u boji kože na licu osobe. Iako se videosnimka svodi na niz slika, njena klasifikacija je složeniji problem. Mreža stoga ima konvolucijske i rekurzivne slojeve. Konvolucijskim slojevima nastoje se izvući značajke i regije na kojima bi se mogla vidjeti promjena. Korišteni su unaprijed naučeni konvolucijski modeli VGG16 i EfficientNet u sklopu prijenosnog učenja. Svrha rekurzivnog ConvLSTM sloja je uhvatiti i interpretirati promjene tijekom vremena. Za izgradnju modela upotrebljena je biblioteka Keras unutar programskog jezika Python.

Abstract (english)

The aim of this paper is to detect the pulse from a color video. After collecting the videos, a solution using a deep neural network to detect skin discoloration on a person’s face was presented. Although video is equivalent to a series of images, its classification is a more complex problem. Therefore, the neural network has convolutional and recursive layers. Convolutional layers seek to extract features and regions where change could be seen. Pre-trained convolutional models VGG16 and Efficient-Net were addapted to this problem as part of transfer learning. The objective of a recursive ConvLSTM layer is to capture and interpret changes over time. The Keras library within the Python programming language was used to build the model.

• Keywords: duboke neuronske mreže; konvolucijske neuronske mreže; rekurzivne neuronske mreže; prijenosno učenje; VGG16; EfficientNet
• Keywords (english): deep neural networks; convolutional neural networks; recursive neural networks; transfer learning; VGG16; EfficientNet
• Language: croatian
• URN:NBN: urn:nbn:hr:168:494907
• 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: 2023-01-13 13:40:17