This dataset contains OCR-processed data from USA passports converted to the FUNSD format.
The dataset includes information such as 'name', 'page_no', 'width', 'height', 'text', 'bbox', 'segment_bbox', 'segment_id', 'qas', 'image', 'md5sum' extracted from USA passports using OCR technology.
The image is decoded as Base64 and stored in the 'image' column. The data is formatted according to the FUNSD (Form Understanding and Annotation for Document Structure) format, suitable for multimodal training tasks.
The dataset can be used for tasks like named entity recognition (NER), key information extraction, and multimodal model training.
