Possibilities of Feedforward Multilayer Neural Network Classifier as a Detector of Pest Birds in Vineyards

Abstract

In this paper, the application of artificial neural network clasifier to resolve pest birds in agricultural areas as a part of a comprehensive system of protection against vermin is demonstrated. Firstly, the idea of the whole system is outlined. Then, the method of recognition is described, the process of artificial neural network design is illustrated and the classifier is validated using data gathered in the fields. Eventually, the results are compared to similar works.

Convolutional Neural Network for Sound Processing - Study of Deployed Application

“…Apparently, it is necessary to develop a robust, selective and cheap enough detection device in order to implement the statistical approach proposed above. In [8] and [11], the authors proposed a detection unit, which used real-time sound recordings as an input source. The sound samples were preprocessed using common filtration and normalization techniques and then, the relevant features were extracted using Linear Prediction Coding approach (LPC) [9].…”

“…Thus, the the pest bird detection device, based on convolutional neural networks, is proposed and designed in the following sections. The performance of the proposed device is compared to the results presented in [8] and in [11], where LPC approach was implemented.…”