Agricultural Pests Classification Using Deep Convolutional Neural Networks and Transfer Learning

Pest image categorization, a novel method based on machine learning algorithms, is used to detect and categorise insects that cause damage to crops in the field of agroforestry. Systems that use a mix of image processing and machine learning to categorise pest pictures in fields may discover pests early, allowing for swift action to be taken to avert crop loss that are vulnerable. Classifying agricultural pests is difficult since there are so many different kinds of pests with such different behaviours and looks. The Deep Convolutional Neural Networks (CNNs) we presented, when paired with Transfer Learning, have proven successful in a variety of picture classification applications, including those of agricultural pests. Therefore, the purpose of this project is to create an exchange-learning-based photo classification model capable of recognising insects from a dataset culled from Kaggle. Good classification performance was shown for recognising pests using our suggested EfficientNetV2 model, with an accuracy rate of 88%. We can easily measure the accuracy using different metrics like training and validation loss and accuracy graphs.