engleski

The use of artificial neural networks as a tool for detection of lepidopteran apple pests

One of the most important apple pests are insects from the order of butterflies (Lepidoptera) - the codling moth (Cydia pomonella (Linnaeus, 1758), whose larvae feed on apple fruit, making it unusable for the market, and the pear leaf blister moth (Leucoptera maifoliella (O. Costa, 1836), whose larvae develop in apple leaves. Classical pest monitoring methods are unreliable and time- consuming, resulting in greater damage to apple production. The use of artificial neural networks (ANN) has recently shown great potential for pest monitoring. Therefore, the aim of this paper is to present ANNs as a pest detection tool that can be used for automatic monitoring of Lepidoptera apple pests (Figure 1). Looking at the examples from the literature where ANNs are used for apple pest monitoring and comparing their accuracy, it can be seen that all effective models for codling moth detection have an accuracy of over 90% in most cases compared to manual counts by human experts. The model for pear leaf blister moth is still pending, but since it belongs to the same order as codling moth (Lepidoptera), the model accuracy should also be high. In addition, ANNs have been used to detect damage to leaves caused by the pear leaf blister moth and also achieved high accuracy. Further development of ANNs for detection and monitoring of important apple pests is certain. Thus, this study reveals an unexplored potential for the use of ANNs in monitoring apple pests from the order Lepidoptera. Therefore, this work advocates more efficient and rapid monitoring that allows for targeted and effective pest control without unnecessary insecticide treatments and thus without negative agricultural impacts on the environment and human health.

ANNs ; Cydia pomonella (Linnaeus, 1758) ; Leucoptera maifoliella (O. Costa, 1836) ; pest monitoring ; smart agriculture

nije evidentirano