use case · basics
The whole idea, plainly. Box one line of printed text; the model reads it back. The characters appear as they're generated — that's the decoder actually running, not a typing animation.
Drag on the image to select a line.
← Back to TrOCR · Next: Practical →