Old-Handwritten OCR Comparison
I am a technical consultant, who has been working as part of Version 1's Innovation Labs since August 2019. For this article, I will compare the famous OCR engine Tesseract, with cloud services such as Azure Read, Google Vision OCR and AWS Textract for handwritten text.
What is OCR?
Optical Character Recognition (OCR) is a conversion of images containing typed, printed or handwritten text in the form of a scanned document into machine-readable text. OCR is widely used for documents such as invoices, bank statements, business cards, receipts, etc.
Options in the Market
OCR services are widely available in the market. To build a solution that can reliably extract handwritten text from a document, a reliable OCR engine is needed.
Cloud solutions from major cloud service providers such as Azure, AWS, and Google can also be used. Google has sponsored the development of Tesseract, a popular OCR engine which has been available since 2006. Pytesseract is an OCR package which is available for Python, being a wrapper for Google’s Tessaract OCR Engine. It is also useful as a stand-alone invocation script to Tesseract.
Handwritten OCR Comparision
Accuracy
This table indicates the number of Text correctly identified by each solution.
Execution Time
This table indicates the time required (in seconds) to OCR each test case.
Note:
• Tesseract was executed locally and therefore there was no network latency
• A 300-dpi setting was used for all solution, which is considered as standard
Data retention and usage policy
Another important point to note is the Data retention and usage policy. Because our customers always care about the data.
Deployed on-premise and managed by our organisation, no data leaves your network.
Azure
As per azure privacy and terms of usage.
“Your images are automatically deleted after processing. Microsoft does not train on your data to enhance the underlying models.”
Please visit https://aws.amazon.com/textract/faqs/ for more information.
“Currently, Google does not use the content you send to train and improve our Google Vision features such as its machine perception model.”
Please visit https://aws.amazon.com/textract/faqs/ for more information.
AWS
“Amazon Textract may store and use document and image inputs processed by the service solely to provide and maintain the service and to improve and develop the quality of Amazon Textract and other Amazon machine-learning/artificial-intelligence technologies. Use of your content is necessary for the continuous improvement of your Amazon Textract customer experience, including the development and training of related technologies. We do not use any personally identifiable information that may be contained in your content to target products, services or marketing to you or your end users. Your trust, privacy, and the security of your content are our highest priority and we implement appropriate and sophisticated technical and physical controls, including encryption at rest and in transit, designed to prevent unauthorized access to, or disclosure of, your content and ensure that our use complies with our commitments to you.
Please visit https://aws.amazon.com/textract/faqs/ for more information.
Final Note:
Azure provides good accuracy with decent execution time. But finally, all OCR service provides does not provide 100% accuracy.
The Innovation Labs has been in action since 2018 and has had many success stories in the form of successful collaborative Proof-Of-Values (PoVs). We are keen on engaging more without current and new customers to demonstrate how the latest technologies can add value to their business. To find out more about Innovation at Version 1, visit us here.
About the Author
Satheesh Mohan is a Solution Architect here at Version 1.