Does anyone know of an open source Intelligent Character Recognition (ICR) library? There seems to be plenty out there for OCR but I am focusing on recognizing handwriting and would love it if there was an open source solution out there.
OCRopus claims to do handwriting recognition. It's available under the Apache License 2.0.
OCRopus does not appear to be ICR-ready at this time. From an OCRopus developer forum message on Dec 18, 2009.
Pretty much all thee old code for handwriting recognition is there and it's now accesssible from Python; it should be fairly easy to script a handwriting recognizer. However, making it perform well will require a lot of training and tweaks, plus careful language modeling. Depending on the nature of the handwritten material, it may also be necessary to integrate additional recognizers (HMM, convolutional nnet).
I was looking for exactly the same as you. However, that (and other similar comments) were enough to convince me that there would be a great deal of work required to get OCRopus ready for prime time ICR. I could not in all conscience recommend it to my client right now.
Also, I see that there has not been a new release of OCRopus in over a year. Hardly reassuring.
来源：https://stackoverflow.com/questions/730594/foss-intelligent-character-recognition-icr