OCR via Web API
I want to send a picture to a API to do the OCR. But after I posted successfully the response told me it doesn't work. The OCR webAPI can't read the body data. The API use the base64 encode for the picture. And when I setup the header to content-type: application/x-www-form-urlencoded,chartset=uft-8, I found in the result the header is still the application/json. How can I configure it?
Thanks,
Xia
Answers
-
@XIA_175253 Can you share why you are using an API action for OCR, instead of using the OCR action? If the API is public, can you also share a link to the API documentation? Thank you!
Images: Optical Character Recognition (OCR)
https://help.catalytic.com/docs/images-optical-character-recognition-ocr/0 -
Hi Jozef,
We are developing a OCR function for Chinese Invoice verification which include the invoice format and Chinese characters.
So we use the local API function from BAIDU who is the one of the biggest AI provider in china market.
Please find the attached document from their portal. And i also attached the link but that's in Chinese.HTTP Method:POST
Request URL: https://aip.baidubce.com/rest/2.0/ocr/v1/general_basicURLparam:
参数 值
access_token combine the API Key,Secret Key to get the access_tokenHeader:
Param value
Content-Type application/x-www-form-urlencodedBody:
Params
Param Must Style
image string - 图像数据,base64编码后进行urlencode,要求base64编码和urlencode后大小不超过4M,最短边至少15px,最长边最大4096px,支持 jpg/jpeg/png/bmp 格式,当image字段存在时url字段失效0 -
@XIA_175253 We are currently checking in to address this issue as soon as possible. I will share a new update as soon as this gets addressed. In the meantime, a different body configuration should do something similar. If your body is like
foo=url-encoded-bar
it should now look like{ "foo": "bar"}
. Does this body configuration meet your use case as well?0 -
Hi Jozef,
Yes, the body is meet the case. like {"image":img}0