I would like to write a simple project that can recognise different written Chinese characters. Now I have been searching various web pages and other sources for a sufficient database to use.

While there are several databases offering free data to use, most of the data is not labelled. The closest to a success I have come was from this web page: http://www.hcii-lab.net/data/onHCCTestdataset/onHCCTest.html

On this web page I found "labels" and "data", however the labels seemed to far outnumber the data. So I can't be sure, if the files are related.

It would already be enough for me, if I had a data points and a labels files I could process (somehow) - the processing aspect of the data shouldn't be the difficult part. It also does not have to be "much" data - I don't have to have enough data so that any and every Chinese characters can identified.

In fact my expectation for the project aren't high - it's intended to be just a project I do for fun over my summer holidays. It's more important to me, that I can try and utilise Machine Learning algorithm for a hobby (Chinese) and in the end have something that works.


1 Answer 1


This resource has a link to a bunch of online and off line databases for handwritten letters.

This offline database seems to be a big one, hosted in Japan by Electrotechnical Laboratory has 1.2 million samples from Chinese, Japanese and other languages.

Nakagawa Labs has another online database, but it's not openly available.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.