Tesseract OCR vs Firebase ML Kit

[babadoctor]

Hello

 

I would like to develop an image recognition app for iOS (and later android) that can recognize the text of a street sign.

 

How do I know which library to use?

 

https://pub.dev/packages/tesseract_ocr

This is supposedly developed by google and used in google translate.

https://developers.google.com/ml-kit/vision/text-recognition/ios

This one is made for recognizing street signs.

 

However... I don't know what exactly is the best one.

 

Are these two the same?

 

 

This is what they describe it as on the firebase website:

You can use ML Kit to recognize text in images, such as the text of a street sign.

 

 

Does anyone on this website have any experience with OCR in ios or android apps?

 

 

[igormp]

I'd go with ML-kit just because it seems easier to implement.

 

For actual performance, you'd need to test both by yourself since I found no easy way to know if they used the same dataset for training.

[babadoctor]

4 minutes ago, igormp said:

I'd go with ML-kit just because it seems easier to implement.

 

For actual performance, you'd need to test both by yourself since I found no easy way to know if they used the same dataset for training.

Alright, thanks!

[Franck]

I used tesseract many times in the past. Probably the most performant OCR I have ever worked with. I was unaware of a IOS, Android implementation. I just recall it was C++