UDOP: Document Question Answering

UDOP, a unified model for document classification, layout parsing and visual question answering by Microsoft.

Try it in the Widget Center

Click this url to try this widget and copy the Pro Config template.

Usage

<TODO: enter description here, and remove useless inputs>

Input Parameters

Name	Type	Description	Default	Required
image	string	image url for udop to understand.	default_url
prompt	string	The text prompt in natural language. You can use `Question answering. xxxxxx` to ask question about the input image	Question answering. In which date is the report made?

Output Parameters

Name	Type	Description	File Type
response	string	The answer given. According to the image and prompt

Output Example

success

{
   "response": "9/3/92"  // example input image=https://replicate.delivery/pbxt/KZczgVh1gAp7xfPP79GdZoGdKEoekcLPiqSqE6bEgM5pThGD/example_1.jpg  prompt=Question answering. In which date is the report made?
}

fail

{
   "results": "<the example results of this widget>"
}

Detailed Guidelines