Skip to content

ldscr/cnn_mnist_demo_with_GUI

Repository files navigation

cnn_mnist_demo_with_GUI

handwriting on a GUI, and predict the digit be wrote.

first build the cnn--->train it with MNIST dataset to get the model--->write a python program reading a img and useing the trained model to predict---->write a simple GUI to use the python program created by the fommer one.

first build the cnn---->cnn_mnist.py followed the tutorial of the tensorflow website https://www.tensorflow.org/tutorials/layers

get the model------->saved in the directory model_data if don't want to waste time trainning the network. can use the trained model directly read img and use trained model to predict------->cnn_mnist_predict_with_img.py the GUI -------->cnn_mnist_predict_with_GUI.py

the environment: lsb-release -a : Ubuntu 16.04 (64 bit) nvidia-smi: Driver Version:390.25 docker: 1.0.1 tensorflow container: 1.5.0

About

handwriting on a GUI, and predict the digit be wrote.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages