PREDICTION MODEL OF CITRUS MAIN PESTS AND DISEASES BASED ON NLP TECHNOLOGY

Chuxin Chen, Shanqin Wang^*

College of Resources & Environment, Huazhong Agricultural University, Wuhan, 430070, China

In view of the poor predictiveeffect of traditional citrus main pests and diseases prediction model which cannot accurately predict citrus fruit pests and diseases and improve the level of citrus production information, a citrus main pests and diseases prediction model was designed based on NLP technology. Firstly, image acquisition, image cropping, image background processing and target point selection were used to realize the preprocessing of citrus pest images. Afterward, the retrieval process of citrus main pests and diseases was designed based on NLP. Next, a number of points were selected in the harm-like region and the normal region from several representative images, and the R, G, B values and their mean values of these sample points were counted.Then, the maximum gray difference between the hazardous region and the normal region wasdesigned and solved for the linear programming objective function, and the RGB component prediction model wasconstructed. Finally, 960 citrus fruit images were processed usingthreshold segmentation method, and the validity of the design model was verified. Compared with other traditional models, the designed model has the largest gray difference between the damaged area and the normal area, so it has the highest recognition accuracy and the lowest false detection rate and omission rate. This indicates that NLP technology can be used to identify citrus fruit pests and diseases.