GENERATING ARTICLE POLYGONS WITHIN NEWSPAPER IMAGES FOR EXTRACTING ACTIONABLE DATA
The present disclosure is directed toward systems, methods, and non-transitory computer-readable media for generating and providing actionable data from newspaper articles identified and segmented from digital newspaper images. For example, the disclosed systems segment articles of a newspaper image by using specially designed models to generate polygons defining article boundaries within the newspaper image. In some cases, the disclosed systems further determine article text from a polygon of an article for additional processing to determine an article topic, determine an article type, predict entity names within the article, and/or predict a locality associated with the article.
1 . A computer-implemented method comprising:
receiving a newspaper image depicting digitized newspaper content;
detecting an article within the digitized newspaper content of the newspaper image utilizing an article prediction model;
generating, utilizing the article prediction model, a polygon defining boundaries of the article within the newspaper image; and
determining text within the polygon utilizing an optical character recognition model.
2 . The computer-implemented method of claim 1 , further comprising:
extracting text features from the text within the polygon utilizing a domain-adapted information extraction model; and
predicting, from the text features, entity names within the text of the polygon.
3 . The computer-implemented method of claim 1 , further comprising generating, using an article locality model, a locality prediction for the article based on the text within the polygon.
4 . The computer-implemented method of claim 1 , wherein detecting the article within the newspaper image comprises:
determining text columns within the newspaper image using a column prediction model; and
aligning the article within one or more of the text columns.
5 . The computer-implemented method of claim 1 , wherein generating the polygon comprises generating an irregularly shaped polygon enclosing portions of multiple text columns within the newspaper image based on the article spanning the multiple text columns.
6 . The computer-implemented method of claim 1 , further comprising:
extracting visual features from pixels of the newspaper image enclosed by the polygon; and
classifying the article into an article topic based on the visual features.
7 . The computer-implemented method of claim 1 , wherein generating the polygon defining the boundaries of the article comprises utilizing a column prediction model repurposed from an architectural building detection model.
8 . A non-transitory computer readable medium storing instructions which, when executed by at least one processor, cause the at least one processor to:
receive a newspaper image depicting digitized newspaper content;
detect an article within the digitized newspaper content of the newspaper image utilizing an article prediction model;
generate, utilizing the article prediction model, a polygon defining boundaries of the article within the newspaper image; and
determine text within the polygon utilizing an optical character recognition model.
9 . The non-transitory computer readable medium of claim 8 , further storing instructions which, when executed by the at least one processor, cause the at least one processor to:
extract text features from the text within the polygon utilizing an information extraction model adapted to a newspaper domain from a free text domain; and
predict, from the text features, name parts for entity names within the text of the polygon.
10 . The non-transitory computer readable medium of claim 8 , further storing instructions which, when executed by the at least one processor, cause the at least one processor to generate, using an article locality model based on the text within the polygon, a locality prediction indicating whether the article is a local article or a non-local article.
11 . The non-transitory computer readable medium of claim 8 , further storing instructions which, when executed by the at least one processor, cause the at least one processor to detect the article within the newspaper image by:
generating, utilizing a column prediction model, a plurality of column predictions for the newspaper image;
determining, from the plurality of column predictions, text columns within the newspaper image according to densities of column predictions at respective locations; and
aligning the article within one or more of the text columns.
12 . The non-transitory computer readable medium of claim 8 , further storing instructions which, when executed by the at least one processor, cause the at least one processor to generate the polygon by generating a rectangular polygon enclosing one or more text columns within the newspaper image including text of the article.
13 . The non-transitory computer readable medium of claim 8 , further storing instructions which, when executed by the at least one processor, cause the at least one processor to:
extract visual features from pixels of the newspaper image enclosed by the polygon; and
classify the article into an article type based on the visual features.
14 . The non-transitory computer readable medium of claim 8 , further storing instructions which, when executed by the at least one processor, cause the at least one processor to generate the polygon defining the boundaries of the article by utilizing a column prediction model repurposed from an architectural building detection model.
15 . A system comprising:
at least one processor; and
a non-transitory computer readable medium storing instructions which, when executed by the at least one processor, cause the system to:
receive a newspaper image depicting digitized newspaper content;
detect an article within the digitized newspaper content of the newspaper image utilizing an article prediction model;
generate, utilizing the article prediction model, a polygon defining boundaries of the article within the newspaper image; and
determine text within the polygon utilizing an optical character recognition model.
16 . The system of claim 15 , further storing instructions which, when executed by the at least one processor, cause the system to:
extract text features from the text within the polygon utilizing an information extraction model adapted to a newspaper domain from a free text domain; and
predict, from the text features, name parts for entity names within the text of the polygon.
17 . The system of claim 15 , further storing instructions which, when executed by the at least one processor, cause the system to generate, using an article locality model based on the text within the polygon, a locality prediction indicating whether the article is a local article or a non-local article.
18 . The system of claim 15 , further storing instructions which, when executed by the at least one processor, cause the system to detect the article within the newspaper image by:
generating, utilizing a column prediction model, a plurality of column predictions for the newspaper image;
determining, from the plurality of column predictions, text columns within the newspaper image according to densities of column predictions at respective locations; and
aligning the article within one or more of the text columns.
19 . The system of claim 15 , further storing instructions which, when executed by the at least one processor, cause the system to generate the polygon by generating a rectangular polygon enclosing one or more text columns within the newspaper image including text of the article.
20 . The system of claim 15 , further storing instructions which, when executed by the at least one processor, cause the system to:
extract visual features from pixels of the newspaper image enclosed by the polygon; and
classify the article into an article topic based on the visual features.