Family tree identification method and system based on deep learning

Abstract

The invention provides a family tree identification method and system based on deep learning. The method is oriented to the field of family tree digitization; based on a deep learning method, a familytree data rapid digitization scheme with two deep learning tasks of target position detection and target classification separated is designed. Firstly, two convolutional neural networks of target position detection and target classification are trained respectively to judge the positions and contents of Chinese characters in a family tree, and finally, a character relationship in the family treeis analyzed through a regular expression to draw a digitized family tree. According to the genealogy identification scheme based on the deep learning, a large amount of manpower identification is reduced and avoided, and the accuracy of a data digitization result is ensured.

Description

technical field [0001] The invention relates to a family tree recognition method and system based on deep learning. Background technique [0002] "The country has its history, the region has its aspirations, and the family has its genealogy." As one of the three systematic historical documents alongside official history and local chronicles, the current digitization of genealogy lags far behind the former two. The digitization of the official history has already been completed, and the local annals are in progress. At present, the digitization of the official history has been completed and entered the field of commercialization, accounting for one-third of the total. The digitization of family trees has not yet begun. [0003] The reason is that there are several difficulties in the digitization of family trees: [0004] 1. The number of genealogy spies in our country is extremely large and expanding, and relying on manpower to extract a large amount of information from ge...

AI Technical Summary

Problems solved by technology

[0004] 1. The number of genealogical spies existing in my country is extremely large and expanding, and relying on manpower to extract a large amount of information from genealogy data consumes manpower and material resources

[0005] 2. There are often a large number of uncommon characters in the family tree, which itself lacks a large number of labeled data sets, and most of the existing deep learning data sets do not contain rare characters, so the existing OCR tools have a lot of problems in recognition. Mistakes

[0006] 3. The records of family trees are structural, and it is difficult to identify the relevance and meaning of the contents of each section

Images