unilm
unilm copied to clipboard
how to get charseg.npy from ocr.txt and image.png
If I want to train with my own dataset, an inevitable problem is how to obtain charseg.npy by processing the image.png and the ocr.txt