Skip to content

ranihorev/Kuzushiji_MNIST

Folders and files

NameName
Last commit message
Last commit date

Latest commit

152ce0e · Dec 14, 2018

History

2 Commits
Dec 14, 2018
Dec 13, 2018
Dec 14, 2018

Repository files navigation

Kuzushiji-MNIST - VGG, ResNet & Capsule Net Classifiers

Kuzushiji-MNIST is a new alternative dataset for the well-known MNIST. The new dataset can serve as a new benchmark system for classification algorithms and can help restoring millions of books from the almost lost Japanese language - Kuzushiji.

The paper describes the dataset, its extraction process and a few use cases and results.
A summary of the paper can be found here.

This code achieves state-of-the-art results (98.9%) on Kuzushiji-MNIST using an ensemble of VGG Network and ResNet and implemented with PyTorch and FastAI (v1).

It also provides a Kuzushiji-MNIST Dataset class to help others use the new dataset.

About

Kuzushiji-MNIST - Capsule Network, VGG and ResNet classifiers for the "lost" Japanese language dataset

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published