Orchard complex road segmentation method based on lightweight semantic segmentation algorithm

Abstract

The invention relates to the technical field of orchard road analysis, and particularly discloses an orchard complex road segmentation method based on a lightweight semantic segmentation algorithm, and the method comprises the steps: obtaining a data set in a preset range, and determining an image recognition model based on the data set; acquiring a to-be-detected image in real time, and performing feature extraction on the to-be-detected image based on the image recognition model to obtain a feature map; performing convolution operation on the feature map based on a trained depth separable convolution model; wherein the number of output channels of the feature map is unchanged; and determining a path and a position relation of the path in the to-be-detected image according to the convolution operation result. Compared with other models, the method has the advantages that the identification effect is not much different, the requirement of orchard complex path information extraction is met, the algorithm operation efficiency is effectively improved, the requirement of an agricultural robot for the calculation power required by visual identification in an orchard environment is reduced, and the requirement of subsequent visual navigation for real-time performance is met.

Description

technical field [0001] The invention relates to the technical field of orchard road analysis, in particular to a method for segmenting complex orchard roads based on a lightweight semantic segmentation algorithm. Background technique [0002] Hills and mountains are the main landforms in the south of the Five Ridges in my country, and they are also the production areas of important economic crops such as forests and fruits. With the progress of urbanization, the labor force in rural areas is decreasing, and the problems of labor shortage and high cost in hilly and mountainous agricultural production are becoming more and more prominent. Promoting the mechanization and intelligence of agricultural production is one of the effective ways to solve this problem. Autonomous navigation of agricultural robots is an important step in the realization of automated operations. In addition to using satellites for positioning and path planning, it is especially important to use vision s...

AI Technical Summary

Problems solved by technology

[0003] However, most of the existing deep learning algorithms are deployed and applied through transfer learning, and the algorithm models trained through public datasets are transferred to specific fields for application. In this process, more consideration is given to the accuracy of the algorithm rather than Ignoring the real-time nature, and they have extremely high requirements on computing resources, they are not suitable for real-time detection process, and it is difficult to be directly applied to intelligent robots

Images