UV Light Pest Control System Enhanced By Deep Learning

Pest detection and management are critical for sustainable agriculture and assured food security. Traditional UV light traps have proved very effective in trapping these pests, but their efficiency could be improved by the use of selective light traps that may be able to attract certain pest species. This paper describes a new approach in pest detection and management using selective light traps with deep learning analysis. A powerful deep learning model, YOLOv5, is trained in this paper with a custom pest data set and used for real-time pest detection. The model has an accuracy rate of $\mathbf{9 1. 3 \%}$. The model identifies the presence of pests reliably in agricultural environments. This paper conducts a comprehensive survey of the literature to determine optimal wavelengths that would attract the target species of pests. The information obtained is then utilized to design selective light traps that would emit a wavelength, to which identified pest preferences are tailored. This can be very promising for the solutions related to inspection and management of pests in agriculture, hence applicable to sustainable agriculture and food security.