Application of backpropagation method for quality sorting classification system on white dragon fruit (Hylocereus undatus)

Fitri, Zilvanhisna Emka and Baskara, Ari and Silvia, Mega and Madjid, Abdul and Imron, Arizal Mujibtamala Nanda (2021) Application of backpropagation method for quality sorting classification system on white dragon fruit (Hylocereus undatus). IOP Conference Series: Earth and Environmental Science, 672 (1). pp. 1-6. ISSN 1755-1307

	Text (Full Paper) 2.Fitri_2021_IOP_Conf._Ser. _Earth_Environ._Sci._672_012085.pdf - Published Version Restricted to Repository staff only Available under License Creative Commons Attribution Share Alike. Download (512kB)
	Text (Cek Similarity) 2. IOP_Application of Backpropagation Method for Quality Sorting Classification System on White Dragon Fruit (Hylocereus undatus).pdf - Supplemental Material Available under License Creative Commons Attribution Share Alike. Download (4MB)

Abstract

Several problems related to determining the quality of dragon fruit quality are: fruit disease, harvest time selection, sorting process and post-harvest grading. Determination sorting dragon fruit quality by observing the appearance of fruit, fruit smoothness, presence or absence of defects and fruit size. However, this quality determination has disadvantages such as longer sorting time and different perceptions of farmers about the quality of dragon fruit. To solve this problem, we need a sorting system that is able to determine the quality of dragon fruit effectively and efficiently without damaging the dragon fruit. In this study, determining the quality of white dragon fruit using digital image processing techniques and intelligent systems. The output of the digital image processing technique is five morphological features such as area, perimeter, length, diameter and metric. This feature is the input of the backpropagation method so that the quality of white dragon fruit is divided into 3 classes such as class A, class B and class C. The results showed the best network architecture model was 5,8,5,3 with the best testing accuracy rate of 86.67%.

Actions (login required)

View Item