Skip to content

Hiragana Classifier - classify different hiragana characters from handwritten images (taking inspiration from the MNIST dataset)

Notifications You must be signed in to change notification settings

SaabriinMo/AI_Coursework

Repository files navigation

Hiragana Classifier

Hiragana forms one of three part of the Japanese writing system, others being katakana and kanji. Hiragana is known to be the easiest to understand. To help me and my partner understand the handwritten hiragana written by japanese people.

website we used to help code the model:

https://github.com/ItsWajdy/tensorflow-tutorials/blob/master/Basic%20Classification.py

https://github.com/rois-codh/kmnist -> I used the balanced accuracy score to get the accuracy of the models.

Make sure you have the following version installed:

tensorflow == 2.0.0-alpha0 keras == 2.2.4 numpy == 1.18.1 matlibplot == 3.1.3

-- This is what verisons i have installed

You may choose to run the code, please ensure you have both juypter notebooks. Since the code was split, you may need to combine the 2 juypter notbook to get the full picture. Also, you may obtain different results from the time the report was created. Just make sure you keep this in mind.

About

Hiragana Classifier - classify different hiragana characters from handwritten images (taking inspiration from the MNIST dataset)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published