Recognizing Handwritten Numbers with Python and TensorFlow

Level: Real-World

Project Brief

Today we will build a program that recognizes handwritten digits using Python and the TensorFlow machine learning framework.

Please download this NPZ dataset. The dataset contains both training images and testing images. These are small images of handwritten numbers from 0 to 9. Here is an example of an image containing the digit “7”.

Your steps to building the program would be:

1. Import the libraries

2. Load the dataset and preprocess the data

3. Build and compile the model

4. Train the model using the training data

5. Evaluate the model

6. Make predictions using the testing data

If you want to view the NPZ dataset images before starting with the project, you can use the following Python script. It will display the 34th image of the training data. You can change the “33” index in the code to another number to see another image.