Title:
非定型文書から構造化情報を抽出するディープラーニングに基づいた方法及びシステム
Document Type and Number:
Japanese Patent JP7393509
Kind Code:
B2
Abstract:
To provide a deep learning-based method of extracting structured information from an atypical document, which is implemented by at least one processor of a computing device.SOLUTION: A deep learning-based method of extracting structured information from an atypical document is provided, the method comprising receiving an input image, extracting a token sequence indicative of a structure of the input image from the input image using a deep learning-based encoder-decoder model, and converting the token sequence into structured information.SELECTED DRAWING: Figure 1
More Like This:
Inventors:
Kim Ki Wook
Hong Tae-gyu
Lim Moonbin
Park Seung-hyun
Hong Tae-gyu
Lim Moonbin
Park Seung-hyun
Application Number:
JP2022189082A
Publication Date:
December 06, 2023
Filing Date:
November 28, 2022
Export Citation:
Assignee:
NAVER Corporation
International Classes:
G06V30/416; G06F40/137; G06F40/14; G06N3/0455; G06N3/0464; G06N3/0475; G06T7/00; G06V10/82; G06V30/412
Domestic Patent References:
JP2020166658A | ||||
JP2020140495A | ||||
JP2021081933A | ||||
JP2019082814A |
Foreign References:
US20190050640 |
Other References:
M. Pourreza et al., "Persian OCR with Cascaded Convolutional Neural Networks Supported by Language Model",2020 10th International Conference on Computer and Knowledge Engineering (ICCKE),米国,IEEE,2020年10月29日,pp.227-232
Attorney, Agent or Firm:
Tadashige Ito
Tadahiko Ito
Osamu Miyazaki
Tadahiko Ito
Osamu Miyazaki
Previous Patent: Channel design support device and channel design support program
Next Patent: Enhancement concept of parallel encoding function
Next Patent: Enhancement concept of parallel encoding function