Which text-spotting model is the most suitable for getting texts from images with a standard height and a very high width? Typical models work with heights and widths of a similar size, and feeding them by image portions causes trouble with texts spotted on more portions.