Extracting structured information from document images

1. A method, comprising:

receiving, by a processing device, a document image;

detecting a tabular structure within the document image;

identifying a plurality of rows of the tabular structure, wherein each row of the plurality of rows comprises plurality of lines;

for each row of the plurality of rows, identifying a respective plurality of lines comprised by the row;

detecting, in each line of the plurality of lines, a set of fields corresponding to a respective set of field types; and

extracting information from the set of fields.

2. The method of claim 1 , wherein the tabular structure is provided by a table.

3. The method of claim 1 , further comprising:

identifying vertical boundaries of the tabular structure.

4. The method of claim 1 , wherein identifying the plurality of rows further comprises:

identifying a plurality of lines of the tabular structure; and

classifying each line of the plurality of lines.

5. The method of claim 1 , wherein identifying the plurality of rows further comprises:

identifying a plurality of lines of the tabular structure; and

clustering the plurality of lines into a plurality of clusters.

6. The method of claim 1 , wherein identifying the set of field types further comprises:

determining, for each line of the one or more lines, a corresponding line type derived from a corresponding set of field types of the one or more fields comprised by the line.

7. The method of claim 1 , further comprising:

detecting a multi-line field by grouping two or more single-line field portions.

8. A system comprising:

a memory; and

a processing device operatively coupled to the memory, the processing device configured to:

receive a document image;

detect a tabular structure within the document image;

identify a plurality of rows of the tabular structure, wherein each row of the plurality of rows comprises plurality of lines;

for each row of the plurality of rows, identify a respective plurality of lines comprised by the row;

identify a set of field types of one or more fields comprised by each line of the one or more lines comprised by the respective row;

detect, in each line of the plurality of lines, a set of fields corresponding to a respective set of field types; and

extract information from the set of fields.

9. The system of claim 8 , wherein the processing device is further configured to:

identify vertical boundaries of the tabular structure.

10. The system of claim 8 , wherein identifying the plurality of rows further comprises:

identifying a plurality of lines of the tabular structure; and

classifying each line of the plurality of lines.

11. The system of claim 8 , wherein identifying the plurality of rows further comprises:

identifying a plurality of lines of the tabular structure; and

clustering the plurality of lines into a plurality of clusters.

12. The system of claim 8 , wherein identifying the set of field types further comprises:

determining, for each line of the one or more lines, a corresponding line type derived from a corresponding set of field types of the one or more fields comprised by the line.

13. The system of claim 8 , wherein the processing device is further configured to:

detecting a multi-line field by grouping two or more single-line field portions.

14. The system of claim 8 , wherein the processing device is further configured to:

train a table detection classifier for detecting one or more tabular structures within a document;

train a vertical boundary detection classifier for identifying vertical boundaries of the one or more tabular structures;

train a row detection classifier for determining a row layout of the one or more tabular structures;

train a line classifier for extracting information from the fields of the one or more tabular structures;

train a field detection module for detecting fields of the one or more tabular structures.

15. A non-transitory computer-readable storage medium including executable instructions that, when executed by a computing system, cause the computing system to:

receive a document image;

detect a tabular structure within the document image;

identify a plurality of rows of the tabular structure, wherein each row of the plurality of rows comprises plurality of lines;

for each row of the plurality of rows, identify a respective plurality of lines comprised by the row;

identify a set of field types of one or more fields comprised by each line of the one or more plurality of lines comprised by the respective row;

detect, in each line of the plurality of lines, a set of fields corresponding to a respective set of field types; and

extract information from the set of fields.

16. The non-transitory computer-readable storage medium of claim 15 , wherein the tabular structure is provided by a table.

17. The non-transitory computer-readable storage medium of claim 15 , further comprising executable instructions that, when executed by the computing system, cause the computing system to:

identifying vertical boundaries of the tabular structure.

18. The non-transitory computer-readable storage medium of claim 15 , wherein identifying the plurality of rows further comprises:

identifying a plurality of lines of the tabular structure; and

classifying each line of the plurality of lines.

19. The non-transitory computer-readable storage medium of claim 15 , wherein identifying the plurality of rows further comprises:

identifying a plurality of lines of the tabular structure; and

clustering the plurality of lines into a plurality of clusters.

20. The non-transitory computer-readable storage medium of claim 15 , wherein identifying the set of field types further comprises:

determining, for each line of the one or more lines, a corresponding line type derived from a corresponding set of field types of the one or more fields comprised by the line.