Image Recognition - Extract text data from uploaded photo

Question

Hi,

I'm looking to build some software which will be able to read a photo of a travel ticket and extract various parts of data for validation. The tickets could take various formats so I'm anticipating needing to train the system to learn those different formats, but I'm not sure where to start with it all. Is the Azure image recognition software capable of accurately recognising and extracting data in this way? Are there any samples/examples of this process in action. 

For example, our system would require users to upload tickets as evidence of travel. We would then want the software to read the ticket they've uploaded and extract the price they paid, the origin and destination of their journey, the ticket reference number etc. 

It feels like this *should* be possible, but I need some guidance as to the best way to achieve this, with modern technology. Can anyone help give me some pointers?

Thanks for your help.

James

Answer 1

Hi,

Sure, it is possible! You can extract text/ hand writing with our Optical Character Recognition. Let me share some document you can refer to first.

https://docs.microsoft.com/en-us/azure/cognitive-services/computer-vision/concept-extracting-text-ocr

https://docs.microsoft.com/en-us/azure/cognitive-services/computer-vision/concept-recognizing-text

Please let me know if you have further issues.

Regards,

Yutong

Answer 2

Hi Yutong,

Thanks for those links. Those pages are very basic and don't go into much detail about more advanced text extraction, and training the software to recognise different layouts of data in an image.

Do you know of a code sample which perhaps reads photos of tickets or receipts and extracts data such as price and date?

Thank you,

James