Ask HN: How to Train Tesseract for Handwriting Recognition of Characters?

3 points by ganeshdeshmukh 7y ago ↗ HN

I can't follow steps in the documentation as here https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract-4.00, I am using tesseract in real-time to recognize OCR printed text/ Optical Character Recognition. But I wanted to extend it further to Handwritten character recognition. currently, the new version of Tesseract doesn't recognize HCR. but it says we can do it after training dataset? I know there are many other ways for the same, but I have to learn Neural-Networks for it, and I would need a faster machine. my GitHub repo is "https://github.com/ganesh-deshmukh/cvkeyboard"

I tried using tensorflow, but I can't pass image correctly so it gave wrong prediction, as this repo 'https://github.com/ganesh-deshmukh/Google-colab-hcr/blob/master/demoHCR.ipynb'

what would be a simple approach for same?

1 comment

[ 3.3 ms ] story [ 15.5 ms ] thread

Adamantcheese 7y ago ↗

From the FAQ: "Can I use Tesseract for handwriting recognition?

You can, but it won’t work very well, as Tesseract is designed for printed text. Look for projects focussed on handwriting recognition."

To which I found https://github.com/githubharald/SimpleHTR, which has a number of references that you should look at.