Handwritten Recognition

  • Last Post 02 October 2017
Christine Lee posted this 28 September 2017


Is there a handwritten recognition function in real time recognition SDK? If not, is there a way to incorporate our own neural network with the SDK to do handwritten recognition?

Cheers :)

Order By: Standard | Newest | Votes
Oksana S. posted this 29 September 2017

The current version of RTR SDK doesn't support a handwritten or handprinted recognition.

About your second question, I am not sure that it is necessary, because RTR SDK accepts as input either a video (a series of frames), or a single image, and returns the output as a plain text. This product does not do something additional except OCR. If you have your own tools for the handwritten recognition, then you do not need to use our SDK.

Another possible solution can be using our online OCR service – ABBYY Cloud OCR SDK. As it is a cloud service and all processing is performed on our servers, this product is independent from the development environment. The only mandatory condition for using it is a stable Internet connection for transmission of the images, the recognition settings to the service and getting the recognition results. But please note that Cloud OCR SDK allows to recognize handwritten text only if the characters are separated and written as single characters (so called "handprinted text"). If this variant is interesting for you, please refer to the How to Recognize Text Fields article for an overview of possible applications. To try the described in this article method, you can our C# GUI Sample > the Field-level tab.

Christine Lee posted this 30 September 2017

Thanks for the reply!

I have looked into ABBYYY Cloud OCR already, but it's not really suitable for the solution I'm developing that requires offline or lack of internet connection.

>If you have your own tools for the handwritten recognition, then you do not need to use our SDK.

As I am in a research group working on handwritten recognition, we wanted to utilize ABBYY RTR SDK for the real-time recognition part and add on our in-house neural network with it. I am just wondering if there were any way to combine these two together at the moment - or if there are any resources on how to develop simple real-time OCRs, it would be much appreciated!



Oksana S. posted this 02 October 2017

Theoretically, you can combine RTR SDK, but our mobile technologies do not support handwritten recognition at all and so far we do not plan to add this possibility.