Detekcija karakterističnih točaka lica dubokim konvolucijskim modelima s primjenom u deidentifikaciji lica

master's thesis

• Title (english): Detection of Facial Keypoints by Deep Convolutional Models with Application to Face De-identification
• Author: Glorija Volarević
• Mentor: Tomislav Hrkać (mentor)
• Committee member: Tomislav Hrkać (predsjednik povjerenstva)
• Committee member: Alan Jović (član povjerenstva)
• Committee member: Zoran Kalafatić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad bavi se detekcijom karakterističnih točaka lica i deidentifikacijom snimljenih osoba pomoću konvolucijskih neuronskih mreža. Rad opisuje osnovne pojmove i principe dubokih i konvolucijskih neuronskih mreža te programsko ostvarenje sustava za detekciju i deidentifikaciju lica koji je temeljen na arhitekturi konvolucijske neuronske mreže. Točke pronađenog lica uspoređuju se s točkama lica u bazi za zamjenu lica iz koje se zatim izvlači lice koje ima najsličniju pozu i izraz kako bi se ostvarila što veća prirodnost deidentificirane slike. Delaunayovom triangulacijom lice se podijeli na trokute te se odgovarajući trokuti slike iz baze preslikavaju na trokute slike koja se želi deidentificirati.

Abstract (english)

This work is about facial keypoints detection and de-identification of recorded people using convolutional neural networks. The work describes basic terms and principles of deep and convolutional neural networks and the program implementation of a detection and de-identification system based on convolutional neural network architecture. Once a face is detected, its facial keypoints are compared to facial keypoints of faces in the face replacement database from which a face with the most similar position and expression is chosen. Using Delaunay’s triangulation, the face is divided into triangles which are then pasted onto the corresponding triangles of the face we want to replace.

• Keywords: detekcija; regresija; deidentifikacija; duboko učenje; konvolucijske neuronske mreže
• Keywords (english): detection; regression; de-identification; deep learning; convolutional neural networks.
• Language: croatian
• URN:NBN: urn:nbn:hr:168:344697
• 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: 2020-01-07 22:55:26