Improved U-net semantic segmentation model construction method and method and system for tea tender shoot recognition and picking point positioning

Abstract

The invention discloses an improved U-net semantic segmentation model construction method and a tea tender shoot identification and picking point positioning method and system, and semi-supervised learning is realized by constructing an improved U-net semantic segmentation network structure and training the improved U-net semantic segmentation network by using a tagged tender shoot data set and a non-tagged tender shoot data set. Tea tender shoot pictures are shot in real time, the tea tender shoot pictures are preprocessed and then input into a trained network model for feature extraction and enhanced feature extraction to obtain a semantic segmentation prediction result, and an improved head prediction network relates to an MSA module to achieve multi-scale channel attention; and finally, according to a semantic segmentation result, positioning a tender shoot picking point coordinate, and realizing real-time identification and picking point positioning of the tea tender shoots based on the improved U-net semantic segmentation model.

Description

technical field [0001] The invention belongs to the field of intelligent tea picking, and specifically relates to a method for constructing an improved U-net semantic segmentation model, a tea sprout identification and picking point positioning method and a picking system based on the improved U-net semantic segmentation model. Background technique [0002] Due to the popularization of tea picking mechanization, the tea picking efficiency has been significantly improved, but the current mechanical tea picking equipment can only pick bulk tea roughly, and cannot selectively pick famous tea. At present, the picking of famous tea still relies on a large number of labors. With the outflow of young people and the aging of the population, the intelligent picking of famous tea has become a research hot spot. rapid development. On the one hand, based on the traditional algorithm, the color of the tea leaves is distinguished to identify the tea sprouts, and the skeleton of the sprou...

AI Technical Summary

Problems solved by technology

[0003] However, the above technologies still have the following defects: 1. The identification and location of tea buds based on traditional algorithms are greatly affected by environmental factors, have low generalization ability, and are difficult to adjust parameters

Not suitable for tea picking in real environment

2. Although the identification and positioning of tea buds based on deep learning is mature, it requires a large number of labeled data sets, the cost of labeling is high, and the efficiency of tea picking is low

Images