PestNet: An End-to-End Deep Learning Approach for Large-Scale Multi-Class Pest Detection and Classification

Aim:

To detect and classify large-scale multi-class pest using Convolution Neural Network

Existing System:

Presently, distinguishing insects in crop fields mainly relies on manual classification, but this is an extremely time-consuming and expensive process. Compared with previous classifiers such as k-nearest neighbors and linear discriminate analysis (LDA), support vector machine (SVM) was proposed with Haar-like features to classify insects and obtained a poor performance than the Convolutional Neural Network.

Proposed System:

Our PestNet consists of three stages: pest feature extraction, pest regions search and pest prediction. In PestNet, the input image is firstly fed into a CNN backbone to extract feature maps, where CSA module is proposed for feature enhancement. Then we fuse RPN and PSSM for providing pest regions and pest prediction respectively. During the prediction phase, Contextual RoIs are presented as contextual information to improve detection accuracy