AI for Acoustic Early Detection of the Red Palm Weevil

In this paper, we present a design of an AI model for the early detection of existence of Red Palm Weevil (RPW) larvae in palm trees to avoid the complete devastation of the yield. The proposed methodology is based on recording the RPW sounds during different activity states then extracts the important features in the sounds by applying MFCC technique. The recorded sounds has been transformed to spectrogram and an ANN has been applied to automatically detect if RPW larvae exist or not. We achieved through the proposed strategy 99.2% accuracy using Alexnet.