Available from version 0.1.11. The Windows version is updated by the Microsoft Store. On a Mac, download the new version from Releases.
What it is
PP-OCRv6 is a text recognition model published by PaddlePaddle, the team behind PaddleOCR. Local OCR includes its small version.
- Languages: Chinese, English, Japanese, and Latin-script languages such as French and German
- Best for: printed text
- Size: about 30 MB — much smaller than the other engines, so it downloads quickly
- Speed: an English page of 47 lines took about 1.2 seconds in our test
- Vertical text: lines are ordered right to left
- Line positions: boxes are drawn on the image
How to use it
- Open Settings with the gear icon at the top right
- In the list of engines, press “Download” next to “PP-OCRv6 (print, multilingual)”
- Choose it in the engine menu at the top, and read
For pre-modern books, PaddleOCR-VL is better
We read 28 pages of woodblock-printed classical Chinese books with both PP-OCRv6 and PaddleOCR-VL. PaddleOCR-VL read the main text more accurately. We found no character that only PP-OCRv6 got right. PP-OCRv6 also picks up text outside the main text, such as the ruler and the library label in the photo.
For pre-modern books, keep using PaddleOCR-VL or NDL古典籍OCR Lite. PP-OCRv6 suits printed and Western-language material when you want results quickly.
If unsure, use “Compare” to try both on your own material. The video in the other news item shows two engines reading the same page, with the differences highlighted.