Western silverware stamp identification method based on deep learning

Abstract

The invention provides a Western silverware stamp identification method based on deep learning, and belongs to the technical field of image recognition. A deep neural network is trained by a manuallylabeled sample. The neural network obtains judgment of producing area and year information by learning stamping characteristics, and the style and the condition of a silverware are directly and comprehensively known and judged.

Description

technical field [0001] The present invention relates to image recognition technology, in particular to a deep learning-based Western silverware stamp recognition method. Background technique [0002] European silverware has an ancient history, and early silverware was mostly used for sacrificial activities and for the upper class. With the innovation of art forms during the Renaissance and the overall improvement of the application of craftsmanship and art appreciation in the Victorian period, a large number of silverware with ingenious ideas, complicated craftsmanship, and various techniques such as hammering, engraving, embossing, hollowing out, and inlaying have survived. Western old silverware has attracted a large number of collectors because of its unique historical value and investment value. Sotheby's and Christie's auction companies hold special antique silverware sales every year. [0003] Since ancient times, most of the handmade silverware in China have not been...

AI Technical Summary

Problems solved by technology

[0005] Most Western silver wares are stamped with the producer's logo, year of production, name of the wares, reasons for making the wares, etc. These stamps are often so small that they are difficult to recognize with the naked eye, and different logos are extremely difficult to identify. It is necessary to use a magnifying glass and stamp illustrations to slowly search for identification

Relying on pure manual recognition has disadvantages such as low efficiency, low accuracy, and long time-consuming

Images