A tool for annotating image datasets, allowing the creation of key-value pairs(json) for each image, with OCR assistanced (PaddleOCR based).
It suitable for creating datasets of Donut/LayoutLMv3(or other e2e models).
Left Mouse to zoom/pan, Right Mouse to draw a box and grab text in it, Esc to clean draw box