Image recognition model training and analysis method based on PaddleOCR

The invention discloses an image recognition model training and analysis method based on PaddleOCR. The method comprises the following steps that a, an external user accesses a deep learning OCR system; b, the deep learning OCR system obtains model configuration and configures data according to para...

Full description

Saved in:
Bibliographic Details
Main Authors CHEN WUCONG, CHEN JINFENG, DENG YAOLONG, ZHONG JIANCHEN, LIANG PEISEN, LI WENHAO
Format Patent
LanguageChinese
English
Published 29.07.2022
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:The invention discloses an image recognition model training and analysis method based on PaddleOCR. The method comprises the following steps that a, an external user accesses a deep learning OCR system; b, the deep learning OCR system obtains model configuration and configures data according to parameters transmitted by an external user; c, obtaining a configuration data object list according to the type code; d, circularly calling OCR (Optical Character Recognition) according to the number of configuration data objects to obtain a recognition result; e, summarizing recognition results, and screening out an optimal result; f, performing post-processing operation according to requirements; and g, finally, uniformly packaging the return object and carrying out return processing. The method can be widely applied to various fields of government affairs, tax affairs, finance and the like, the functions of identifying government affairs materials, identifying tax ticket numbers, identifying financial bank cards and
Bibliography:Application Number: CN202210506508