Abstract
Red palm weevil (RPW) is a harmful pest that has wiped out many palm plantations worldwide. Early detection of RPW is difficult, especially on large plantations. Here, we report on combining fiber-optic distributed acoustic sensing (DAS) and machine learning to detect weevil larvae less than three weeks old, in a controlled environment. In particular, we use the temporal and spectral data provided by a fiber-optic DAS system to train a convolutional neural network (CNN), which distinguishes "healthy" and "infested" signals with a classification accuracy higher than 97%. Additionally, a rigorous machine learning classification approach is introduced to improve the false alarm performance metric by >20%.