The purpose of this project is to use TensorFlow's Neural Network to analyze hand-written digit images and predict the digit or the class of the input image.
We use the following methods for the analysis:
- import the training and test sets from the MNIST database,
- preprocess and prepare the training and test sets for the model,
- create and compile the deep neural network model,
- train the model and run the predictions,
- visualize results.
- Data Source: MNIST Database
- Software: Python 3.8, TensorFlow 2.3.1, Jupyter Notebook 6.
Link to code NN Image Classification
The model created has an accuracy of 97%.
Below is a snippet of a sample of hand-written digit with the predicted classification from our model.