Lightweight solanaceae disease identification method based on SE-Inception

Abstract

The invention discloses a lightweight solanaceae disease recognition method based on an SE-Inception structure, and belongs to the technical field of image classification, and the method comprises thefollowing steps: establishing a common solanaceae disease data set, and carrying out the manual marking; inputting a solanaceae disease data set; enhancing the image and expanding a data set; dividing the data set into a training set, a verification set and a test set; after data is input, designing parallel convolution layers, performing convolution by using convolution kernels of different scales, and adding a BN layer behind each convolution layer; establishing a lightweight network; training a training set in the data set; and testing and outputting the test data set by using the trainedmodel. The composition comprises five types of diseases; a lightweight network architecture is designed, and parameter optimization is continuously carried out in the training process. Experimental results show that the method has relatively high detection precision, relatively high training speed and relatively small model volume, and technical feasibility is provided for deployment of mobile terminals and hardware.

Description

technical field [0001] The invention belongs to the technical field of image classification, and in particular relates to a light-weight solanaceae disease identification method based on SE-Inception. Background technique [0002] Solanaceous crops such as tomato, eggplant, wolfberry, potato, pepper, etc. are widely cultivated crops with high economic value. And disease is a very big factor that influences the crop yield of Solanaceae, and a large amount of crops are encroached by different plant diseases every year, cause a large amount of losses. Accurate detection and identification of plant diseases is a key element affecting plant production and key to successful farming. The main steps of traditional crop disease recognition are to use image processing technology to preprocess crop disease images, extract some specific features, and use classifiers to classify the extracted features, so as to realize the classification and recognition of crop diseases . Wang Liwei, ...

AI Technical Summary

Problems solved by technology

The relatively mature deep learning-based object recognition methods in the current academia have achieved good results on public data sets, but agricultural data has its own special features and cannot be directly transferred and applied

In addition, most of the better recognition network models in the current academic world are relatively large, which cannot meet the production and application requirements of the actual mobile terminal.

Images