Unable to parse <" "> with OCR api

I have a picture with multiple names. One person has a name like this: <“Tinus”>

Whatever i do, it will always come back empty. As if these are html escaped. So the result will always be empty.

As a test i added lines of text and fed it to the OCR with <“thisismytext”> and it will also not parse it, it will come back as empty again. Why does it at least not come back with “thisismytext” ?

Thanks.

Did you test with engine 1 and 2?

As you said, this should work fine. => Please post the image here, then we can test it.

Number 18 is not correct.

Using engine 2.

@admin presume it dropped out of scope ;p

'\n50m ago\t\r' + '\nGroteGreut\t38m ago\t\r' + '\nzo het begin is er weer\t\r' + '\nGewoon luie Erik\tnow\t\r' + '\nSEND\t\r' +

Same deal here too. If you have something with 1111 in it, the whole line is now gone.

Hi, I tested it. In my test all works fine works fine with engine2:

Postman API test:

Overlay:
18

1111 number ocr in the second image works fine, too:

Presume you are using just the default settings. I was also using OCR engine 2. Have a friend that also used a similar setup and had issues with the ‘1111’ in a row. I’ll double check the settings and see if i missed something somewhere.

Yes, I used the default settings, as used by our online ocr front page.

But note that this page uses scale=true as default settings. In the API call, you must add scale=true, since the default is scale=false

See also: Online test result varies from the API tests

I did the same test as you did, it did do it better with OCRengine2. Turns out the ocrengine=2 was added in the wrong section in the call in my javascript…

The text does get a better match but the [ ] around the numbers do have more false returns then ocrengine. Pretty darn unlucky. Ocrengine1 does all [xx] numbers correct but some text not ok where as ocrengine 2 does do all text just fine but the numbers sometimes incorrect.

Gues i’ll have to make some regex around it somehow.

Thanks for verifying though!