US 12,260,632 B2
Systems and methods for image based content capture and extraction utilizing deep learning neural network and bounding box detection training techniques
Arnaud Gilles Flament, Sunnyvale, CA (US); Christopher Dale Lund, San Diego, CA (US); Guillaume Bernard Serge Koch, San Jose, CA (US); and Denis Eric Goupil, Sunnyvale, CA (US)
Assigned to Open Text Corporation, Waterloo (CA)
Filed by Open Text Corporation, Waterloo (CA)
Filed on Dec. 19, 2023, as Appl. No. 18/545,885.
Application 18/545,885 is a continuation of application No. 17/147,324, filed on Jan. 12, 2021, granted, now 11,893,777.
Application 17/147,324 is a continuation of application No. 16/035,307, filed on Jul. 13, 2018, granted, now 10,902,252.
Claims priority of provisional application 62/533,576, filed on Jul. 17, 2017.
Prior Publication US 2024/0135700 A1, Apr. 25, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06V 10/82 (2022.01); G06F 18/214 (2023.01); G06F 18/2413 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06V 30/14 (2022.01); G06V 30/18 (2022.01); G06V 30/19 (2022.01); G06V 30/413 (2022.01); G06V 30/414 (2022.01); G06V 30/10 (2022.01)
CPC G06V 10/82 (2022.01) [G06F 18/214 (2023.01); G06F 18/24143 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06V 30/1444 (2022.01); G06V 30/18057 (2022.01); G06V 30/19173 (2022.01); G06V 30/413 (2022.01); G06V 30/414 (2022.01); G06V 30/10 (2022.01)] 20 Claims
OG exemplary drawing
 
1. An image recognition system, comprising:
a processor; and
a non-transitory computer-readable medium; and
stored instructions translatable by the processor to perform:
obtaining a plurality of document components, wherein each document component represents a corresponding feature type;
dynamically generating a plurality of images of simulated documents, each image being generated from a corresponding subset of the plurality of document components;
providing the plurality of images as inputs to a convolutional neural network and providing, for each image, the corresponding feature types of the document components from which the image was generated as expected outputs; and
training the convolutional neural network to recognize in the images the feature types of the corresponding document components.