Piper nigrum Classification using Artificial Bee Colony Algorithm
| dc.contributor.author | Siti Hazemah binti Hassan | |
| dc.date.accessioned | 2026-10-19T03:21:56Z | |
| dc.date.issued | 2018 | |
| dc.description | The use of bio-inspired optimisation techniques has expanded in recent years. In this research, a bio-inspired optimisation technique is proposed to solve one of the challenges that Malaysian Pepper Board (MPB) is currently facing when classifying mass production of pepper, known scientifically as Piper nigrum. This research explores the Artificial Bee Colony (ABC) algorithm for classification of Piper nigrum according to the standard grades that were set by the MPB based on the ISO standard. ABC is a metaheuristic algorithm inspired by the collective behaviour of bees, which has been used successfully to solve complex problems of optimisation. With this as an advantage, the use of ABC is proposed with expectation of having to be able to effectively classify Piper nigrum according to its own grades. A conceptual approach for the proposed simulation is constructed in order to make abstract distinctions in the concept, organise ideas and structure it into a model. The approach ideally having to firstly pre-process the training and testing data sets, then use the data sets to learn and model a structure that in the end will classify each data sets into a grade. This model consist of the utilisation process of feature selection and a fusion approach, which applied into the ABC algorithm. Using these feature selection and fusion approach in ABC, the feature subset is obtained and can be used to predict the unknown grade of the testing data sets. Then, finally a classifier known as K-Nearest Neighbour (KNN) will be used to provide an outcome of how well the data sets were accurately classified based on feature subset obtained earlier. This KNN classifier is also used to validate how accurate the testing data sets classified based on the best feature selected. In the experiment, data sets are being tested in two scenarios. First scenario consists of testing against a data set that is combined into one large batch called Ungrouped Data Set. Second scenario consists of testing against a data set that is being grouped into its five groups. In both of the scenarios, all testing data set is used iv to validate the model derived from the training data set. From this research, it is concluded that Piper nigrum Classification using ABC is able to provide 89.70% accuracy when tested with ungrouped data set and 100% accuracy when tested with grouped data set. Piper nigrum Classification using ABC was effectively able to out-perform other feature selection techniques available in the Waikato Environment for Knowledge Analysis (WEKA) by 14.61% when compared to ungrouped data set and by 24.91% when compared to grouped data set. Piper nigrum Classification using ABC also contributed to a better pepper grading process compared to current colour sorter machine as the machine was only able to sort the peppers based on colour feature only. The Piper nigrum Classification using ABC was also the first computerised pepper grading process that able to sort white pepper berries based on colour, shape and texture features. | |
| dc.identifier.uri | https://scholarhub.unimas.my/handle/123456789/3146 | |
| dc.language.iso | English | |
| dc.publisher | UNIVERSITI MALAYSIA SARAWAK | |
| dc.relation.ispartofseries | Faculty of Computer Science and Information Technology | |
| dc.subject | Optimisation algorithm, artificial bee colony (ABC), image processing, feature selection, image classification, k-nearest neighbour (KNN) | |
| dc.title | Piper nigrum Classification using Artificial Bee Colony Algorithm | |
| dc.type | Masters |