Emotional Recognition VIA Subtle Human Interaction (Facial Expression) Using Convolutional Neural Network

Koay, Xian Hong (2018) Emotional Recognition VIA Subtle Human Interaction (Facial Expression) Using Convolutional Neural Network. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text (24 Pages) Emotional Recognition VIA Subtle Human Interaction (Facial Expression) Using Convolutional Neural Network.pdf - Submitted Version Download (304kB)

Abstract

Machine learning have been booming recently. Especially with the existence of better GPU and CPU (which allow better computational power), deep learning, a machine learning technology that need the large computational power, can now effectively process the large amount of data that we currently have. This allow wide array of applications, one of which can be apply to machine vision. Using the technology of machine learning, this project set out to create a program that learn to recognize emotion via subtle human interaction using the intricacy of facial expression. One would be able to implement said program in daily life using real time evaluation, allowing it to detect early stage of depression, provide companionship to the elderly and aid in service industry. Using Convolutional Neural Network and deep learning algorithm, this project managed to create a program that recognize human expression and identify the emotion in real time. The result of this project shows certain amount of promises, the project achieved a validation accuracy of 40% to 50%. Despite that, the real time evaluation using the trained model is quite accurate most of the time, the accuracy drops drastically for human with spectacles, however. With more time and resources, it is possible to improve this project further than what currently is.

Actions (login required)

View Item