Orchard fruit tree region segmentation method and system based on deformable convolutional neural network

Abstract

The invention discloses an orchard fruit tree region segmentation method and system based on a deformable convolutional neural network, and belongs to the field of intelligent agriculture. The orchardfruit tree region segmentation method comprises the steps: for depth images and color images of the same fruit tree area in an orchard, extracting depth features of the depth image; extracting an initial color feature of the color image by using a deformable convolutional neural network; utilizing the deep features and the initial color features to learn together to obtain offset parameters of the deformable convolutional neural network; and obtaining a feature amplification coefficient of the deformable convolutional neural network through deep feature learning, obtaining a feature model, collecting a color image of an orchard, extracting color features of the color image through the feature model, and segmenting the color image through the color features to obtain a fruit tree region. According to the orchard fruit tree region segmentation method, the complex form of the fruit tree can be better modeled, so that the fruit tree area is more accurately segmented to reduce the waste ofpesticides, and the pollution of the pesticides to the land is reduced, and the orchard fruit tree region segmentation method is of great significance to the implementation of intelligent agriculturein China.

Description

technical field [0001] The invention belongs to the field of intelligent agriculture, and more specifically relates to a method and system for segmenting orchard fruit tree regions based on a deformable convolutional neural network. Background technique [0002] The research on the segmentation method of orchard fruit trees can help the precise application of pesticides in orchards, which can not only reduce the waste of pesticides, but also reduce the pollution of land caused by pesticides, which is of great significance to the implementation of intelligent agriculture in my country. [0003] The traditional orchard fruit tree area segmentation algorithm based on image features uses artificially designed features such as color and texture to segment fruit tree areas, but its segmentation accuracy is seriously disturbed by factors such as uneven illumination and weeds in the background. The orchard fruit tree area segmentation algorithm based on the 3D model is segmented thr...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the above defects or improvement needs of the prior art, the present invention provides a method and system for orchard fruit tree area segmentation based on deformable convolutional neural network, thus solving the problem that existing models in the prior art cannot adapt to the complex shape of fruit trees and the problem of planting of fruit trees Technical problems of mis-segmentation of interval areas, waste of pesticides, and increase of environmental pollution

Images