Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/522322
Title: Facial expression recognition by machine and deep learning methods
Researcher: Sumalakshmi, C H
Guide(s): Vasuki, P
Keywords: Computer Science
Computer Science Information Systems
Engineering and Technology
Expression Recognition
human-computer interaction
nonverbal communication
University: Anna University
Completed Date: 2023
Abstract: Facial expressions play a key role in nonverbal communication, by newlineconveying about one s behaviour, feelings, motivations, or intentions. Facial newlineExpression Recognition (FER) is the method of labelling the expressions on face newlineimages into categories like anger, fear, surprise, sadness, happy, disgust and so newlineon. FER is a vital part of human-computer interaction that enables computers to newlinecomprehend facial expressions based on human thought. It has numerous newlineapplications in the field of security, marketing, e-learning, robotics, mental newlinehealthcare purposes, surveillance and so on. The most serious problem of the newlineexisting facial expression recognition methods is its poor prediction accuracy. newlineTherefore, the objective of this research is to enhance the recognition accuracy. newlineFacial expression recognition involves steps like facial feature newlineextraction, feature optimization, and expression classification. Human brain newlinegathers information about an expressive face, not just from its physical newlinecharacteristics such as wrinkles, but also from its geometrical aspects, such as the newlinedegree to which the eyes, mouth, and brows are drawn together. There are newlinegenerally two types of features in a face image. First category is the appearance newlinefeature or texture-based feature that extracts the local region characteristics from newlineimages. The second category is geometric feature which are the features extracted newlinebased on the shape and locations of facial components. Thus, Facial feature newlineextraction is more effective while using highly discriminative appearance as well newlineas geometric features. The human brain decodes and decorrelates face information newlineby processing both the geometrical and appearance facial features at distinct newlinelevels. Hence, instead of just concatenating them, these features have to be fused newlinein the right way to obtain the correct representation. The high dimensional feature newlinerepresentation may contain lot of irrelevant and redundant features which needs newlineto reduced. newline
Pagination: xix,142p.
URI: http://hdl.handle.net/10603/522322
Appears in Departments:Faculty of Information and Communication Engineering

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File23.93 kBAdobe PDFView/Open
02_prelim_pages.pdf3.09 MBAdobe PDFView/Open
03_contents.pdf31.76 kBAdobe PDFView/Open
04_abstracts.pdf125.83 kBAdobe PDFView/Open
05_chapter1.pdf256.85 kBAdobe PDFView/Open
06_chapter2.pdf220.33 kBAdobe PDFView/Open
07_chapter3.pdf495.56 kBAdobe PDFView/Open
08_chapter4.pdf240.17 kBAdobe PDFView/Open
09_chapter5.pdf3.25 MBAdobe PDFView/Open
10_chapter6.pdf795.34 kBAdobe PDFView/Open
11_annexures.pdf151.94 kBAdobe PDFView/Open
80_recommendation.pdf63.23 kBAdobe PDFView/Open
Show full item record


Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).

Altmetric Badge: