I am using the vision "type": "DOCUMENT_TEXT_DETECTION" to detect handwritten forms/documents. The response JSON contains the bounding box information of
