A Wood Texture Classification Method Based on Migration Learning

Abstract

The invention relates to the interdisciplinary technical fields of deep learning, computer vision technology, and wood texture detection methods, and discloses a wood texture classification method based on transfer learning, which obtains training data by acquiring wood image data and preprocessing it; constructs a method for recognizing wood texture The migration learning model of the category; use the pre-trained convolution model to obtain the convolution model package for recognizing the wood texture category after training; according to the actual production environment, the eigenvalues ​​​​calculated by each model are combined with the recognition method by optimizing the calculation time. The results are judged to obtain the final classification. The present invention is designed according to the actual production situation. It is good in terms of reliability and practicability, and can realize automatic classification of wood texture in production. The classification is accurate and saves time and manpower. cost.

Description

technical field [0001] The invention relates to the interdisciplinary technical fields of deep learning, computer vision technology, and wood texture detection methods, in particular to a wood texture classification method based on migration learning. Background technique [0002] The traditional classification of wood texture mainly relies on the experience of workers, which is highly subjective and inefficient. Therefore, it is urgent to introduce automatic means to realize efficient and accurate classification of wood texture. At present, there is no research on wood texture recognition and classification using deep learning. This patent introduces a wood texture recognition and classification method based on deep learning convolutional neural network. Since there is no strict definition of classification standards for each category in wood texture, so in There are huge challenges in partitioning datasets and training convolutional model parameters. [0003] The convolu...

AI Technical Summary

Problems solved by technology

[0003] The convolutional neural network model based on deep learning can automatically extract high-order features to achieve accurate classification of high-similarity images. However, the convolutional neural network has many parameters. And requires a large amount of labeled data, which brings limitations to the implementation of deep learning models

Images