Extracting $

[cole-w-kelley]

From a few brief tests, it appears this project extracts $ very poorly. It almost always extracts them as 3s or 5s.

[HT0710]

The text recognition component of this project was trained exclusively on a Vietnamese dataset, therefore its failure to recognize the '$' symbol is expected.
Unfortunately, I'm uncertain if an update is possible due to lost data access.
Maybe, I will create an improved version during my free time.

[cole-w-kelley]

What data access was lost? Would it be possible to retrain the model?