tesseract
tesseract copied to clipboard
Music Symbol Not Recognised
Environment
- Tesseract Version: 4.1.1
- Platform: Windows 64-bit
Current Behavior:
Music symbol is detected as letter 'O'
Expected Behavior:
To recognise the symbol? It used to work before...
I am also having the same problem as mentioned in another issue here: https://github.com/SubtitleEdit/subtitleedit/issues/6076
It used to work before...
@greentrancer, which version worked? Which Tesseract models are used to recognize musical symbols?
@greentrancer, @ScribbleGhost, could you please add an example image and the tesseract
command line which reproduces the issue?
@stweil Are you saying that Tesseract should be able to recognize musical notes?
For me, neither Tesseract 3.02 nor Tesseract 5.1.0 has been able to recognize music notes. I am not familiar with the command line for Tesseract. I have only used SubtitleEdit. There is no way to display logging for Tesseract in SubtitleEdit, so I am not sure what is happening in the background.
Here is an example image:
No, I asked which version worked because I was surprised to hear that it "used to work before". Personally I don't know a Tesseract model which could recognize musical notes. Maybe SubtitleEdit has its own algorithm to detect them.
That makes sense to me. I think I will stick to SubtitleEdit's nOCR as I have no idea how to train Tesseract and can't be bothered to be honest. nOCR takes a little while to train, but the interface is easy.
Had similar problem with 5.1.0 working very poorly on music notes in subtitle edit. Switched "Engine Mode" to "Original Tesseract only (can detect italic)" setting and it's recognizing most everything now. Tons better.