Advanced Pest Identification: An Efficient Deep Learning Approach Using VGG Networks
K. LakshmiNadh, Divvela Chandu Venkateswara Guptha, J Shiva Sai, Kandula Rajesh, Sireesha Moturi, Yaragani Neelima, Dodda Venkata Reddy
Abstract
Accurate pest identification is crucial for both effective pest management and crop protection. Pests must be found early in order to minimise damage and guarantee crop security. Conventional techniques typically entail visual examination and professional involvement, which might be time-consuming and susceptible to errors by humans. On the other hand, deep learning-powered high-performance systems can now more accurately identify pests thanks to developments in computer vision. In this work, we employed the Keras-based deep learning models VGG16 and VGG19 to construct a passive pest detection system. We greatly improved the efficacy of these models in identifying pest species by using strategies such data augmentation, model optimization, and modification of validated models. The VGG16 model produced an amazing accuracy rate of 99.8% and VGG19 model produced an accuracy of 96.8 % in our testing.