The purpose of this work is to develop a recognition system of 
the Beetle Colorado Potato (Leptinotarsa decemlineata) pest. 
We work with images and we use them in the system based on 
neural networks. We use a collection of 25 images with 
different position of beetles, also with varying numbers of 
beetles. We calculate the brightness, contrast and  contour 
orientation histograms for feature extraction of every image as 
input to the Random Subspace Classifier (RSC) neural 
classifier. We describe in this paper the structure, algorithms, 
and the initial results of the RSC classifier testing.