Hello, everyone. I am still learning Label Studio and have a challenge getting pre-detected OCR text to properly appear in the Label Studio UI.
I’m trying to use label studio to annotate document images. Before I load up the images I run an OCR scan and use its results to automate some of the annotations.
I’m trying to include the actual text the OCR read in the data, so I add it as part of the JSON object I send like this:
{
“id”: f"ocr-{i}",
“type”: “labels”,
“value”: {
“x”: x_min,
“y”: y_min,
“width”: width,
“height”: height,
“text”: text
},
“to_name”: “image”,
“from_name”: “label”,
“labels”: [“Text”]
}
It seems to be partially working, because all of the boxes from the OCR read is represented on the images in the UI in label studio, and I can see that the x, y, width, and height values exist. That being said I can’t find where i can view the “text”: text part in the UI and I need that in order to correct some of the text boxes drawn by the OCR read. I also need that data to exist when I pass it all to train a model eventually.
I can see the UI has a “meta” value, but I don’t think that’s relevant here?
Hopefully somebody can help me out. Thank you so much!