OCR Trainer asks for more samples but it actually has more than recommended
1 次查看(过去 30 天)
显示 更早的评论
Dear All,
I am training an OCR for a specific software tool typography. I have 20 samples per character for the training (which is actually double of the minimum recommended). However, when I train the algotithm it starts processing the 200 images but at the end this message appears:
"Unable to tain OCR. Please provide additional training samples for each character. At least 10 samples per character is recommended for training OCR."
What should I do? Does anyone run into the same problem? Could this error be because my images look very similar among them?
Thank you
0 个评论
回答(2 个)
Marcin Wozniak
2021-1-1
Hello,
i solved this problem by merge all pictures in one image. Next step i executed "train button" in OCR trainer and works well.
Try this one:
Marcin
0 个评论
Matteo Bonhomme
2023-3-28
I also got this issue. I added even more sample (at least 100 per character) and it ended up working
0 个评论
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Text Analytics Toolbox 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!