Proceedings

Find matching any: Reset

Add filter to result:

Artificial Neural Network Techniques To Predict Orange Spotting Disease In Oil Palm

S. Liaghat, S. K. Balasundram

Department of Agriculture Technology, Faculty of Agriculture, University Putra Malaysia, 43400 Serdang, Selangor

Large-Scale oil palm plantations require timely detection of disease symptoms to enable effective intervention. Orange spotting is an emerging disease that significantly reduces oil palm productivity. Remote sensing technology offers the means to detect crop biophysical properties, including crop stress, in a cost effective and non destructive manner. In this study, different portable sensors were used to measure spectral reflectance and chlorophyll content of orange-spotted oil palm foliage. Spectral reflectance data and chlorophyll content were obtained from the leaves of healthy (non-infected), mildly infected, moderately-infected and severely-infected oil palm trees.Artificial Neural Network (ANN), a highly simplified nonlinear modeling tool, was used to predict orange spotting disease severity.A multilayer feed-forward neural network trained with an error propagation algorithm was applied. Two training algorithms belonging to two classes have been evaluated: gradient descent algorithm, and Levenberg–Marquardt algorithm. Different ANN architectures, with various hidden neurons were examined for each training algorithm. The root mean squared error (RMSE) and coefficient of determination (R²) were used to choose the best neural network model. The network trained with the quick propagation algorithm (QP) showed the best overall predictive ability.

Keyword: Precision agriculture, orange spotting disease, spectral reflectance, chlorophyll content, neural network

Download paper