Automated Grading of Edible Bird Nest

Koay, Mei Yuan (2018) Automated Grading of Edible Bird Nest. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Koay Mei Yuan.pdf Restricted to Registered users only Download (2MB)

Abstract

Edible bird’s nest (EBN) is one of an expensive animal bio-product produced by swiftlets due to its mineral contents and ability to bring various medicinal benefits. Among the producers of EBN such as Indonesia, Thailand, Philippines and other countries, Malaysia is the third largest producers of EBN. Annual production of EBN in Malaysia may reach billions of ringgit. Despite the painstaking harvesting process, the processing of EBN, which includes the grading and cleaning process consumes high amount of time, cost and labour. The conventional way of grading the EBN is carried out by human expert, where the grading may be subjective and inconsistent among different individuals. The cost to hire an expert for inspection will also increase the overall cost of the EBN. Therefore, this research suggested that automated technology can be implemented in the system to provide aid in the grading process by using machine vision and Adaptive Neuro Fuzzy Inference System (ANFIS). In the current EBN industry, automated grading system is not yet extensively implemented thus it is a challenge to design such system to increase the production rate and reduce the labour cost. The objective of this research is to design an intelligent system to automatically and accurately classify the EBN into different grades consistently. The colour images of the raw EBN were pre-processed to extract the features such as the area of EBN, the area of impurities, the colour of EBN and the curvature of EBN. Classification of EBN based on the features was then done by using ANFIS and the results obtained were compared with the k-Nearest Neighbour (kNN) which serves as a baseline for the accuracy. An accuracy of approximately 92.80% is achieved with the EBN dataset with the proposed method.