Method for actively positioning and focusing crop diseases and insect pests based on convolutional neural network

Abstract

The invention relates to a method for actively positioning and focusing crop diseases and insect pests based on a convolutional neural network. The invention belongs to the technical field of crop pest location positioning and focusing. The objective of the invention is that in combination of the latest convolutional neural network technology, through researching a large number of pest and diseasedamage part images, pest and disease damage part characteristic parameters are obtained, and a pest and disease damage part feature algorithm is formed; and the pest and disease damage part can be judged only by substituting image parameters into the algorithm, so that the pest and disease damage part is automatically positioned and focused in the photographing process, the pest and disease damage part is marked, background interference is weakened, and the pest and disease damage identification accuracy and efficiency are improved. A convolutional neural network technology is adopted, so that the recognition focusing response speed is high; through recognition and classification of each pixel of the image, a plurality of pest and disease damage parts on one image can be recognized at thesame time, and recognition is more comprehensive; automatic positioning focusing self-learning can be assisted by manually selecting and adjusting positioning each time, and characteristic parametersare updated, so that positioning focusing is more and more accurate.

Description

technical field [0001] The invention relates to a method for actively locating and focusing crop diseases and insect pests based on a convolutional neural network, which belongs to the technical field of positioning and focusing of crop diseases and insect pests. Background technique [0002] At present, there are quite a lot of recognition technologies in the field of pest recognition, and the image recognition methods they use include object-to-object image comparison recognition, SVM statistical vector machine method and convolutional neural network technology. The first two images Recognition technology is relatively backward, and image recognition using convolutional neural network technology is also based on the recognition of the entire image. In a relatively perfect recognition environment, pests and diseases can be accurately identified. When put into the actual agricultural planting environment The recognition rate is often reduced due to the interference of the co...

AI Technical Summary

Problems solved by technology

[0002] At present, there are quite a lot of recognition technologies in the field of pest recognition, and the image recognition methods they use include object-to-object image comparison recognition, SVM statistical vector machine method and convolutional neural network technology. The first two images Recognition technology is relatively backward, and image recognition using convolutional neural network technology is also based on the recognition of the entire image. In a relatively perfect recognition environment, pests and diseases can be accurately identified. When put into the actual agricultural planting environment The recognition rate is often reduced due to the interference of the complex and changeable natural environment in the field. In order to ensure the recognition rate, the complex image background is processed and recognized after the image is uploaded, which will increase the processing pressure on the server side, making image recognition more difficult for the hardware terminal. The configuration requirements are increased, and the recognition response time will be greatly lengthened and the recognition efficiency will be reduced.

Images