A multilayer perceptron approach for Ficus carica (fig) ripening classification / Mohd Ikmal Fitri Maruzuki … [et al.]

The ripening stage is a stage where the fruit is ready to be harvested. During ripening, pectin activity is observed to trigger parenchyma cell wall middle lamella dissolution of a fruit. Additionally, the ripening stage also affects the changing appearance of the fruit. Thus, this research aims to...

Full description

Saved in:
Bibliographic Details
Main Authors: Maruzuki, Mohd Ikmal Fitri, Shahrin, Aisyah Sakina, Setumin, Samsul, Ramli, Rafidah Aida, Senin, Syahrul Fithry, Mohamed Talib, Mohd Shukri, Osman, Mohamed Syazwan
Format: Article
Language:English
Published: Universiti Teknologi MARA Cawangan Pulau Pinang 2017
Subjects:
Online Access:http://ir.uitm.edu.my/id/eprint/46603/1/66603.pdf
http://ir.uitm.edu.my/id/eprint/46603/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The ripening stage is a stage where the fruit is ready to be harvested. During ripening, pectin activity is observed to trigger parenchyma cell wall middle lamella dissolution of a fruit. Additionally, the ripening stage also affects the changing appearance of the fruit. Thus, this research aims to develop a classification model based on ANN that can predict or classify the ripening stage based on either pectin activity or fruit appearance. The study will focus specifically on Ficus carica (fig). To achieve the objective, the researchers developed two Multilayer Perceptron (MLP) models: figNN and pectinNN. We trained figNN using features extracted from images of figs with different ripening stages, and pectinNN with a set of the statistical value of pectin activity such as weight (W), brix of sugar (BS), extraction yield (EY), and degree of esterification (ED) from 30 figs with varying degree of ripening. From the result of this research, figNN and pectinNN can distinguish the ripening stage based on either the chemical properties or the images. Furthermore, we can also show that the image-based classification is more accurate than the pectin-based classification. For future work, the study of the correlation between pectin and image features is highly encouraged.