Pushing metadata for document information extraction task

[saif-fares]

Hello,

I am working on information extraction task on document images. So basically, data are images and i have to annotate bounding boxes on the image for each entity (name, date, client adress, ...) and also the text values contained in each of these entities' bounding boxes.

Annotating bounding boxes on documents for all entities' is already a very heavy thing to do manually, if in addition i have to write the corresponding text value myself it would be more and more heavy.

So my question : is there any way to push on Labelme OCR results corresponding to each image of my dataset and to have automatically the ocr associated to each bounding box i draw on the document as a form of pre-annotation for the text part (given that an OCR gives as the text but also its position on an image). This way, all i have to do is to draw the object detection bounding box, then only correct the associated pushed ocr text instead of writing it from scratch myself.