Our API crops all image document formats, including image PDFs. It also identifies images with multiple, separate documents and divides them as a multiple-image PDF; cropping is controlled by the crop_document parameter in API requests (true by default) and is disabled when boost_mode is on.
sample:
Original image:
Cropped image:
Original image with multiple documents:
Cropped multiple-page pdf:
If you wish to only get one document as a response, regardless of the content of the original image, you can set max_pages_to_process!=1; see more.
Known issues:
Overlapping or having documents too close together can cause the model to fail in detecting several distinct documents:
A small gap between documents is recommended to increase the chances of a correct crop:
