Please use this identifier to cite or link to this item:
Title: Development of visual search services using deep learning (Facial expression classification with convolution neural networks)
Authors: Seng, Eugene Rui Hao
Keywords: DRNTU::Engineering
DRNTU::Engineering::Electrical and electronic engineering
Issue Date: 2018
Abstract: Facial expression classification and recognition is a very popular topic amongst researchers nowadays. Being able to identify human emotions would provide a very strong advantage in many contexts like market research or developing social robots. Humans have been gifted this ability to recognize facial expressions with virtually no effort or difficulty, however effective expression recognition by machines still poses a challenge. This is especially so for unconstrained, in the wild datasets. Due to the uncontrolled environments in which the pictures are acquired, the natural setting poses many obstacles for effective classification such as a variance in lighting condition, occlusions and image quality. In this paper, an approach based on Convolutional Neural Networks (CNN) will be employed to tackle this problem space. The trained CNN model will be used to predict the facial expression label of the input image and classify them into one of these seven categories: angry, disgust, fear, happy, neutral, sad and surprise. Transfer learning will be adopted to cope with the lack of training data. The experiments conducted will exemplify the effects of controlled and uncontrolled datasets as well as the effectiveness of different models on facial expression classification in the wild. The datasets involved in the experiments are two frequently used public dataset: Cohn Kanade Extended (CK+) and Labelled Faces in the Wild (LFW).
Schools: School of Electrical and Electronic Engineering 
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
EugeneSeng_U1522850L_FYP Final Report.pdf
  Restricted Access
FYP Report3.49 MBAdobe PDFView/Open

Page view(s)

Updated on Jun 23, 2024

Download(s) 50

Updated on Jun 23, 2024

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.