Automated verification system for document packages

Challenge

One of kt.team's clients, a large logistics company, processes thousands of shipments every day. This is a complex business process that involves working with multi-page packages of paper documents, digitizing them and sending them to several information systems used by the company.

It was decided to develop a system for automated verification of document packages in Python, with the ability to implement machine learning and computer vision technologies.

Solution

We needed to develop a solution that would automate the processing of paper documentation and would be integrated into all the client's business processes and eliminate manual operations at work.

Customer problem

Our clients had an entire department (about 50 people) that manually processed packages of documents.

The main task is to automate processes and relieve staff of more than half of the burden of processing paper and digitized documents.

As part of the project, ICDocs was developed, an intelligent system for automated verification and sorting of document packages developed in Python.

What ICDocs is looking at

ICDocs features

Digitization of paper documents

The operator uploads a batch of documents to the scanning device. ICDocs starts the scanning cycle.

Translating images into text

It translates scanned images into text, determines the type of documents and the number of pages.

Verification of documents

The operator has the opportunity to assess the correctness of data recognition on documents and mark fields that do not correspond to reality.

Preparation of a package of documents

Reads the name of the legal entity, recognizes the order and shipment number. Receives information about counterparties, transactions, and supplies from accounting programs. Checks for page numbering and the number of pages in the document, and combines all pages in the document. Combines documents with a single attribute (invoice number, travel number, counterparty) into one package.

Sending to external programs

If you wish, ICDocs can be integrated into any company's document management system and send packages of documents to accounting programs and repositories, both Russian and international.

Document storage

ICDocs stores backup copies of documents that were processed in it, with the ability to view the history of changes and who made them.

Development results

The automated verification system for packages of documents can work on several sets of documents simultaneously. At the beginning of development, we were able to approach 80% recognition.

Work on the project is still ongoing and we plan to improve recognition algorithms as close as possible to 100%.

‍

ICDocs will be useful if you have:

Lots of paper documents

EDI and paper document management are practiced with various counterparties. We need to bring everything together into a single system to make it easier to control.

Lots of orders every day

Each order is accompanied by a package of documents. It is necessary to promptly check incoming packages for correctness and check, sort them and submit them for further work.

Several legal entities

Several agreements have been concluded with partners for different legal entities; it is necessary to check that the packages for each agreement are complete and that each document is filled out correctly (the legal entity, seal, signature correspond to those specified in the agreement).

Multiple contracts

Several agreements have been concluded with each of the partners, which vary the specifics of cooperation (different delivery providers, different payment methods or approaches, etc.)

Lots of suppliers

You work with hundreds and thousands of counterparties. A large package of documents must be checked for each counterparty.

ICDocs is developed in Python

‍

The advantage of projects developed in Python is the ability to implement machine learning and computer vision.

Python is a universal, object-oriented programming language.

Learn more about Python‍