Directional training platycodon grandiflorum impurity identification method based on spectral imaging

Abstract

The invention discloses a directional training platycodon grandiflorum impurity identification method based on spectral imaging, and belongs to the technical field of depth science. The method comprises the following steps of: firstly, constructing a multi-dimensional label, then constructing a model branch structure for each label, respectively carrying out dichotomy training on each dimension of the multi-dimensional label in a model data training process, and then carrying out error tuning on the whole, thereby realizing dichotomy discrimination for different color attributes of platycodon grandiflorum sundries. According to the method, a multi-dimensional label system is introduced, so that the spectral data belong to a plurality of categories, and priori knowledge is introduced from the perspective of category labels, so that the algorithm can be more interpretable in the classification process. According to the method, different attribute tags are adopted to divide the categories of the spectrums, so that the classification result of the spectrums is more interpretable, training can be performed according to corresponding attributes, and the final classification precision of the model is improved; and high-precision identification and classification of impurities with spectral characteristics different from those of platycodon grandiflorum are realized.

Description

technical field [0001] The invention belongs to the technical field of hyperspectral imaging and deep learning, and in particular relates to a method for identifying impurities of Platycodon grandiflorum based on directional training based on spectral imaging. Background technique [0002] my country is a big country in the production and consumption of bellflower, and the processing and sales of bellflower play an important role in the national economy. However, it is difficult to ensure the purity of the raw materials of the bellflower during the planting and transportation process. If the impurities are not cleaned thoroughly, the final product of the bellflower will have low content of active ingredients, poor flavor of the bellflower, and difficulty in maintaining the consistency of the quality of the bellflower. Serious Problem. At present, deep learning is an effective way to solve the impurities contained in the raw materials of Platycodon grandiflorum. Traditional ...

AI Technical Summary

Problems solved by technology

At present, deep learning is an effective way to solve the impurities in bellflower raw materials. Traditional deep learning uses one-hot encoding as the classification output of training, which certainly keeps the distance between the categories in the vector space the same, but it also makes The prediction results of the model are even less interpretable

Images