DESIGN AND DEVELOPMENT OF A SCENE RECOGNITION SYSTEM USING NEURAL NETWORK MODELS

Ong, Hui Xin (2020) DESIGN AND DEVELOPMENT OF A SCENE RECOGNITION SYSTEM USING NEURAL NETWORK MODELS. [Final Year Project Report] (Unpublished)

	PDF Ong Hui Xin - 24 pgs.pdf Download (830kB)
	PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3973/3933) Ong Hui Xin ft.pdf Restricted to Registered users only Download (2MB)

Abstract

Scene recognition has become one of the challenging aspects in machine learning. Not only that the performance of a state-of-art scene recognition system is bad, but it also requires a powerful computational device in order to carry out tasks. Hence, the central objective of this research is to design a new scene recognition system that performs well, at the same time reduce computational load of a scene recognition system. The biggest modification of the new scene recognition system is that it extracts objects as attributes for a classifier to perform scene classification. It also combines an object detection Convolutional Neural Network (CNN) model and a classifier. The method is simple, as it uses low computational power but also makes the scene recognition system perform well. There are two experiments done in this research to illustrate the performance of the new scene recognition system analysed. From the experiments done to classify different scene classes, it shows good performances of 97.11% and 80.22% of accuracy in classifying two distinct scene classes and three similar scene classes respectively.

Actions (For repository members only: login required)

View Item