Extract information from scanned technical drawings

The better picture

Technical documents such as drawings and manuals are usually presented as scanned images of paper documents and contain no searchable information.

Today, we understand that technical drawings are very structured documents, and capturing data is an important task for automating search.

Example usability

Various methods were used to extract information from these drawings. One of the solutions is OCR (Optical Character Recognition) – full-text recognition of all characters in the document.

A good step forward, but the result of this processing is still an unstructured document.

OCR data classification

EasyData takes a more comprehensive approach to the drawing recognition.
The quality and amount of extracted information have been significantly improved. Our Artificial Intelligence (AI) techniques process the drawings after recognition. This approach provides the foundation for Digital Transformation of any digital archive.

Machine learning

To develop these algorithms, EasyData’s engineers processed and classified millions of technical drawings and ‘informed’ the neural network which values ​​are relevant for data extraction.

Our clients are satisfied with the results of our network. Our algorithms can significantly speed up the document processing by reducing the number of recognition and classification errors. In doing so Artificial Intelligence is learning on each project, which makes it possible to solve increasingly complex challenges.

Daily practice

EasyData can realize an individual project relatively quickly. The implementation speed depends on the complexity and the number of examples that can be used for the training. 

Our algorithms work at the intersection of two fields: computer vision and machine learning. For example, the operator classifies the document and shows the system what data he’s interested in. So our system learns and automatically finds such data on new documents.

High-quality recognition is equally important. The recognized text can be used for classification algorithms to determine to which specific area a given drawing should be attributed. It is even possible with our technology to convert handwritten text into readable text! 

The quality of document recognition, the speed with which we adjust our algorithms to your tasks, as well as the range of problems we solve makes us unique specialists in the world of technical drawing recognition.