)
width, height = image.size

However, this model includes a brand new [~transformers.LayoutLMv2Processor] which can be used to directly
prepare data for the model (including applying OCR under the hood).