A Deep Survey on Techniques Used to Recognize Locust Based on CNN

Abstract

The major threat in agriculture is insect pests and crop disease. Outbreaks and upsurges of insects can cause huge loss to crop production. Locusts are crop devouring pests found in many parts of the world. Recognition of locusts in early stage helps to prevent the spread of locusts by taking appropriate counter-measures and biological control methods. Initially, locusts were classified and recognized manually which is time consuming and requires taxonomic knowledge. Classification and identification of different species of locusts is a difficult task due to their similar appearance. To address this issue, computer vision and artificial intelligence techniques are used to recognize objects of interest in digital image with good performance and accuracy. This review presents convolutional neural network architectures for recognition of locust species. Convolutional neural networks perform automatic feature extraction and learns complex high-level features with less pre-processing required and it were designed to map image data to an output variable and are mainly used for image classification and recognition. This review serves as guidelines for understanding convolutional neural networks.