Hello,
Thanks for your feedback and insight for the product. I will forward your idea to product team and let you know the response.
Regards,
Yutong
This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
Hello,
Just started using the OCR from Cognitive Services and I've found that it's, to be frank, not that good. But it's very easy to use in Logic Apps and it's very inexpensive so I'd like to see how I might be able to improve the readings.
We are taking pictures of a product with a serial number label on it created by the product's manufacturer. The string is long and has a character that looks like a zero throughout the serial number. Sometimes the letter that looks like a zero is adjacent to numbers and other times letters. When this character is next to numbers, OCR interprets it as a zero. When it's next to letters, it's interpreted as an uppercase letter 'O'.
I tried making my own labels with various methods: I have my own small label maker and I took pictures of my screen using different fonts. I even tried using a font that renders the zero with a diagonal line through it and uppercase 'O' as a mostly round circle.
None of the fonts I used (even Word's OCRB font) worked any better and the worst performer was the zero with the line through it. Those characters were interpreted as some kind of UTF-8 character.
My goal here is two fold:
Hello,
Thanks for your feedback and insight for the product. I will forward your idea to product team and let you know the response.
Regards,
Yutong
Thanks for your response. What about the main part of my post to find the best font? Which font is the system trained on currently?