Seed cotton impurity classification and identification method and system

Abstract

The invention relates to a seed cotton impurity classification and identification method and system. The method comprises the following steps that a data set formed by machine-harvested seed cotton impurity-containing images is acquired; a seed cotton impurity classification and identification model is constructed and a loss function of the seed cotton impurity classification and identification model is determined; the seed cotton impurity classification and identification model is a YOLO v4 network model, and the loss function comprises a bounding box regression loss function, a bounding box confidence loss function and a target classification loss function; based on the loss function, darknet53 is used as a framework, the seed cotton impurity classification and identification model is trained by adopting the data set to obtain a trained seed cotton impurity classification and identification model; a to-be-identified seed cotton impurity-containing image is obtained; impurity segmentation is conducted on the to-be-identified seed cotton impurity-containing image, and a segmented impurity image is output; and the segmented impurity image is input into the trained seed cotton impurity classification and identification model, and an impurity classification and identification result is output. According to the seed cotton impurity classification and identification method and system, impurities of seed cotton can be classified and identified, and meanwhile, the identification accuracy is high.

Description

technical field [0001] The invention relates to the field of cotton impurity identification, in particular to a seed cotton impurity classification identification method and system. Background technique [0002] The traditional hand-picked harvesting method of cotton in China seriously restricts the efficiency and economic benefits of cotton picking. With the increase in the proportion of mechanized harvesting year by year, the machine-picked cotton processing technology has gradually developed, and the impurities mixed in cotton have also increased, such as machine-picked seed cotton. Natural impurities such as bell shells, stiff petals, cotton branches, cotton leaves, dust and other impurities in the food. If it is not detected and fed back in time, the working parameters of each equipment on the cotton processing production line cannot be optimized, which will seriously affect the quality of textiles, thereby affecting industrial benefits and exports. [0003] At present...

AI Technical Summary

Problems solved by technology

If it is not detected and fed back in time, the working parameters of each equipment on the cotton processing production line cannot be optimized, which will seriously affect the quality of textiles, thereby affecting industrial benefits and exports

[0003] At present, the research on cotton impurities is mostly focused on lint cotton. After multi-stage cleaning and carding, the impurity content of lint is low, the image background is relatively simple, and the lint layer is thin during detection, so there is no problem of impurity occlusion. Applicable to seed cotton, not to guide cotton processing

Images