Target multi-category real-time segmentation method and system

Abstract

The invention provides a target multi-category real-time segmentation method and system, and the method comprises the steps: inputting an orchard image collected in real time in the driving process of an unmanned vehicle into a trained semantic segmentation network, obtaining a segmented image outputted by the semantic segmentation network, and according to the distribution condition of different color blocks in the segmented image, determining an obstacle distribution state corresponding to the orchard image; wherein the semantic segmentation network is constructed based on a SegNet network. According to the target multi-category real-time segmentation method and system provided by the invention, the semantic segmentation network is constructed based on the improved SegNet network, the method and the system are suitable for navigation of intelligent agricultural machinery and perception of agricultural scenes, and the distribution of obstacles in front of the unmanned vehicle can be determined in real time, so that the unmanned vehicle can be guided to accurately avoid the obstacles. The method effectively overcomes the defects of low intelligent degree, inaccurate recognition and the like in traditional orchard recognition, and can meet the requirements of modern orchard plant protection operation.

Description

technical field [0001] The invention relates to the technical field of agricultural intelligent equipment, in particular to a method and system for real-time segmentation of multiple categories of objects. Background technique [0002] With the continuous improvement and development of intelligent agricultural machinery and equipment, there are higher requirements for the recognition of multi-category objects in orchards. In recent years, the orchard segmentation method based on neural network is an important part of orchard plant protection operations, and it has obvious advantages compared with traditional orchard identification technology. The complex environment of orchards includes obstacles such as pedestrians, utility poles, houses, and vehicles. Accurate identification of plants and various obstacles in orchards has important research significance for safe operations such as automatic driving and human-computer interaction in orchards. [0003] In the field of neura...

AI Technical Summary

Problems solved by technology

If the data set is not expanded, there may be loss of internal data structure and loss of spatial hierarchical information

In addition, although the dilated convolution does not increase the number of parameters, it may cause grid effects, thereby affecting the accuracy of orchard segmentation

Leading to inaccurate identification of orchard segmentation

[0006] 2) In this method, it is difficult to reconstruct the information of small objects, and there are certain limitations

If the recognition speed of the model is slow, it will lead to slow response of the unmanned vehicle during the operation process, and even when it encounters obstacles, it cannot brake and stop in time, and there are certain safety hazards

Images