[HakonAgustsson]

I own this website MyFreeOCR.com you can use it to convert images to text. It supports 100 languages and many different formats (pdf to text, tiff to text, bmp to text, jpeg to text, bmp to pdf, jpeg to pdf, and more. I would like to get feedback and feature request from you. Currently we are looking into developing an OCR API. Would that be useful for you?

[zmix]

> Currently we are looking into developing an OCR API. Would that be useful for you?

Oh yes! Absolutely! However, I only would use it for private purposes, I do not plan to write an application for it, that I'd release. I would do a lot of preprocessing (mostly cutting into boxes, which I would then save in separate bitmaps) and then upload that.

Just these days I am in the process of converting the scans of a recipe-book, that is not available electronically, but which I want to have on my smartphone. There is a PDF online, but that is only a slideshow of bitmap images. It also has ornamental frames on each page, ornaments in text, etc. And as I did not find the results of open source OCR apps to be satisfactory, I started roaming the commercial sector (I still have to evaluate ABBYY Finereader, but that would be very expensive).

Ideally, so I thought, a household would have an account, that is realized via a basic subscription plus pay by volume, and then available as a REST API. The subscription provider would offer a simple user client like https://www.roxyappsdev.com/applications/windows-10-applicat... and a mobile client (think "document scan", with batch and auto deskew/process) but otherwise work (hard ;-)) on the recognition. Now, if hand-writing would also be possible, that would be great!

BTW: What is 'OSD'? And why not some graphics, that explain the different page-selection/-recognition modes?

[vineyardmike]

> Currently we are looking into developing an OCR API. Would that be useful for you?

The above question is different from the below question:

> I own this website MyFreeOCR.com... I would like to get feedback

The feedback I have is (1) website is garbage - too many ads. Clearly not targeting devs who might want an API. (2) OCR API seems like a cool product. Focus on that. Privacy of my data if I call API? How is data trained? What language/alphabet/etc? How do I try it out ? What formats?

Bonus feedback, the best dev API sites are easy to see value proposition and easy to find API key + docs. Count clicks required to get API key + find template cURL copy/paste. And count discrete pages. Even if "coming soon" - show me the critical path today so I can find it later. If you want to start with an "upload a pic" site, still do above, but structure the upload tool as a "try out the API without needed to leave browser" where you upload a photo and can tune other params via web form.

PS. Lots of OCR api's exist, so it could be a cool project but what is your goals here? Do you want to startup? Why your tool and not BigCloud?