https://github.com/chengluyu/naive-digit-recognition
My naïve implementation of a neural network of MNIST dataset.
https://github.com/chengluyu/naive-digit-recognition
Last synced: 3 months ago
JSON representation
My naïve implementation of a neural network of MNIST dataset.
- Host: GitHub
- URL: https://github.com/chengluyu/naive-digit-recognition
- Owner: chengluyu
- Created: 2017-11-06T16:15:50.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2017-11-28T05:41:04.000Z (over 7 years ago)
- Last Synced: 2025-02-08T16:14:35.983Z (4 months ago)
- Language: Python
- Size: 7.52 MB
- Stars: 0
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.rst
Awesome Lists containing this project
README
=======================
Naïve Digit Recognition
=======================
The project consists of two parts: basic task and advanced task.
Basic task is a MNIST digit classifer implemented in neural network.
And the advanced task is a handwriting digits recognizer.
----------
Basic Task
----------
Method: CNN
Accuracy: 99%
Model:
|cnn-model|
-------------
Advanced Task
-------------
The advanced task is aiming to recognize hand-writing digits in photos.
The whole process is divided into many stages, each of which is responsible for a sigle task.
Result:
|exp-result|
Since all of those digits are written by my self, which is totally different from the style of MNIST.
Plus some error in neural network. It only reaches 95% accuracy.
.. |cnn-model| image:: model.png
.. |exp-result| image:: result.png