Motivation for launching the project by the customer:
the need to scale the customer’s business revealed limitations in the current solution: the business model of the solution used to automate document flow using humans had a high cost and low scaling capabilities.
Description of the initial situation:
Project goals:
MIL Team solution:
the use of the team’s existing solutions in the field of detection and recognition of text in images based on neural network models made it possible to quickly implement a solution with the required level of quality.
Results:
Solutions for recognition automation have been built:
Checks and receipts;
Invoice;
Work order;
Contracts.
To build the model we used:
Simulation results:
Customer: Finance, Accounting
Technology stack: TensorFlow, Python, Flask.
the need to scale the customer’s business revealed limitations in the current solution: the business model of the solution used to automate document flow using humans had a high cost and low scaling capabilities.
Description of the initial situation:
- High efficiency of document flow with a minimum number of errors involves storing electronic versions of documents and rapid digitization of printed versions of documents.
- Processing of printed versions of documents involves converting images into an electronic structured representation with high accuracy.
- The business process used for this is not suitable for scaling, since manual labor is used to recognize printed versions of documents.
- For this reason, there were significant risks for the planned business growth.
Project goals:
- creation of a system for recognizing the main fields of documents with a sufficient level of quality
- creation of a tool for generating new document formats.
MIL Team solution:
the use of the team’s existing solutions in the field of detection and recognition of text in images based on neural network models made it possible to quickly implement a solution with the required level of quality.
Results:
Solutions for recognition automation have been built:
Checks and receipts;
Invoice;
Work order;
Contracts.
To build the model we used:
- Marked up document template;
- Document reference fields with their characteristics;
- Set of document images.
Simulation results:
- OCR character extraction model;
- Model for searching text blocks Text Detection;
- Table Detection model;
- Model for constructing an electronic version of a document.
Customer: Finance, Accounting
Technology stack: TensorFlow, Python, Flask.
