Improved YOLOv5 target detection model construction method and method for identifying tea tender shoots and positioning picking points

Abstract

The invention discloses an improved YOLOv5 target detection model construction method and a tea tender shoot identification and picking point positioning method, an improved YOLOv5 target detection model is constructed, an improved CSPDarknet53 structure is designed in a Backbone module, the improved CSPDarknet53 structure is characterized in that a Nonlinear mapping structure is introduced on the basis of a DarkNet53 network, and feature extraction is carried out on a feature map. In addition, an improved YOLOv5 target detection model is utilized to predict the image, and a semantic segmentation map of a target detection and picking area of the tea tender shoots is output; calculating a picking point based on a prediction result; and finally, outputting a tea tender shoot image with picking point information to realize tea tender shoot identification and picking point positioning.

Description

technical field [0001] The invention belongs to the technical field of intelligent tea picking and machine image processing, in particular to an improved YOLOv5 target detection model construction method and a method for tea sprout identification and picking point positioning. Background technique [0002] The intelligent picking of tea buds can significantly improve the efficiency of tea picking, and the identification of buds and the positioning of picking points are the basis for intelligent picking of tea buds. At present, there are two main methods for the identification and detection of tea sprouts and the location of picking points. [0003] One is under the traditional segmentation algorithm, based on the threshold method or the combination of color information features and edge distance to identify the sprouts, and further determine the sprout centroid, which is set as a two-dimensional picking point. This method has a low generalization ability. , the background r...

AI Technical Summary

Problems solved by technology

[0003] One is under the traditional segmentation algorithm, based on the threshold method or combined with color information features and edge distance to identify shoots, and further determine the centroid of the shoots, as a two-dimensional picking point. , The background requirements are high, and the effect of accurate identification and positioning of tea buds in the natural background is not ideal

[0004] The other is an algorithm based on deep learning. The target detection algorithm is used to identify and locate the buds, and the two-dimensional coordinates of the picking point are determined in combination with the skeleton extraction method. However, the target detection algorithm used is basically a horizontal frame, which is not conducive to the tea leaves. Accurate identification and positioning of buds, especially the subsequent skeleton extraction to determine the picking point, greatly lengthened the task time and the accuracy was not high

Images