Compared to v1.2，v1.3 has two improvements: 1、Increased accuracy.I replace the 'latex-ocr' with the 'texify' model which is stronger.But meanwhile,the file size also increases to 800Mb+. 2、Better support for complex input. By introducing a detect model before recognition model, FMatPix now can tackle with multi formulas and multi-line formula better.But pay attention that it'll consume much time.

If downloading via github is slow, try this link: https://pan.xunlei.com/s/VOABKxsc8Ogp-D4lhceXkQCYA1 code:e6fv

20241026 update I've modify some codes of v1.3 which can increase its speed greatly. Now v1.3 allows for better recognition results at a similar speed to v1.2. But there's no easy way to reduce its size(800Mb+) now, after all, it uses a more powerful recognition model than V1.2. Maybe you can try V1.2 first, and if it doesn't meet your needs then download V1.3 !