Oil palm tree counting and disease detection using deep neural network

Citation

Lee, Chee Cheong (2023) Oil palm tree counting and disease detection using deep neural network. PhD thesis, Multimedia University.

Full text not available from this repository.
Official URL: http://erep.mmu.edu.my/

Abstract

Number of oil palm trees and their location is one of the key information for plantation management. This information helps to predict the yield of palm oil, density of the plantation, and they could estimate the amount of fertilizer needed for a plantation area. Apart from the number of trees, oil palm tree disease status is also important for plantation managers. Basel stem rot (BSR), caused by Ganoderma boninense becomes the most severe problem in oil palm plantations. Effective BSR disease detection is crucial to ensure stable palm oil production. Data collection is human intensive and time costly by deploying manpower on site to do the job. It also requires a site expert for disease detection. Thus, it is essential to utilize technology to aid the data collection process. Drone imaging has gained popularity in the agricultural sector in the past few years, thanks to its capability of covering large area inspection, and producing high resolution images. Furthermore, the artificial neural network has achieved outstanding performance in image recognition and classification recently. Integrating these two technologies enables the owner in getting the above information timely and accurately. This work proposed two methods for oil palm tree counting. These methods focus on drone images that have high resolution and are widely deployed in the plantation industry. Analyzing drone images is challenging due to variable drone flying altitudes, resulting in inconsistent tree size in the images. Counting by template matching or fixing sliding windows size method often produces an inaccurate count. A convolutional Neural Network (CNN) with a variable sliding window size technique is proposed to overcome the variable window size problem.

Item Type: Thesis (PhD)
Additional Information: Call No.: QA76.87 .L44 2023
Uncontrolled Keywords: Neural networks (Computer science)
Subjects: Q Science > QA Mathematics > QA71-90 Instruments and machines
Divisions: Faculty of Engineering and Technology (FET)
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 28 Aug 2024 01:27
Last Modified: 28 Aug 2024 01:27
URII: http://shdl.mmu.edu.my/id/eprint/12865

Downloads

Downloads per month over past year

View ItemEdit (login required)