A pearl classification method based on deep learning

Abstract

A pearl classification method based on deep learning, comprising the following steps: 1) obtaining pearl images as sample data; 2) preprocessing the sample data; 3) dividing the sample data into training data and test data; 4) using the training data to train Depth convolutional network; 5) utilize the trained deep convolutional network to extract the features of training data and test data; 6) utilize step 5) to construct the SVM classifier of the feature of the training data extracted, and utilize this SVM classifier to test data sort. The present invention uses multiple views of pearls for deep learning, fully utilizes the self-learning advantages of deep learning, automatically learns good features, and eliminates the cumbersome process of manually extracting features and designing features, and combines SVM to classify pearls, which can effectively It can accurately judge whether the pearl has thread or not.

Description

technical field [0001] The invention relates to the technical fields of image processing and pattern recognition, in particular to a pearl classification method based on deep learning. Background technique [0002] China is a big producer of freshwater pearls, and its output accounts for 95% of the world's output. Among them, Zhuji City, Zhejiang Province is the largest base for freshwater pearl cultivation, processing and sales in my country, and its total output accounts for more than half of the country's total output. It is known as "Chinese Pearl". Town of". Zhuji's freshwater pearl cultivation area has reached 380,000 mu, and it has more than 1,500 pearl processing enterprises. [0003] After collecting a large number of pearls, most pearl enterprises need to manually classify the pearls in order to classify the pearls into different grades. Since manual classification will be affected by many factors, especially in the case of small pearls and large quantities, the c...

AI Technical Summary

Problems solved by technology

[0008] In order to overcome the problems that the existing pearl classification system based on machine vision is difficult to detect and classify the threads of pearls with high precision, and it is not easy to extract the thread features of pearls through manual design features, the present invention provides a pearl classification method based on deep learning , training a large amount of pearl image data through a deep convolutional network, learning the thread characteristics of pearls, and combining SVM to classify pearls, has a high recognition rate, good practicability, and good classification effect

Images