Crop diseases can affect harvested commodity and thus quality of the yield.
When the diseases are estimated early, the yield will increase by taking measures
thanks to farmers. In this paper, daylily leaf images are used as crop sample, derived
from different agricultural sites under expert control and a system is designed in
order to estimate rust diseases on digital daylily leaf images by using Gabor wavelet
based a neural network model. In the first stage, a feature matrix is extracted from
each digital image with using Gabor Wavelet Transform (GWT) and the statistical
parameters are derived from each feature matrix to form a texture feature vector for
each digital image. These parameters are mean, standart deviation and entropy. In
the second stage, GWT based texture feature vectors are applied to different network
structures of neural network model as inputs for classification and the results are
compared in terms of testing performance in order to determine the best network
structure. Daylily leaf images are classified into two (1.Normal, 2.Diseased) groups
and the best average performance is observed as 80.00 % in the (3-25-1) network
structure of neural network model.