This script demonstrates basic image classification with TensorFlow and Keras. The script creates, trains, and evaluates a neural network model is able to predict digits from hand-written digits with a high degree of accuracy.
- Loads the MNIST dataset mnist.npz
- Preprocesses the data including one-hot encoding and normalization
- Constructs a neural network model, utilizing TensorFlow
- Trains the model on the training data
- Evaluates the model's performance on the test set
- Visualizes predictions with sample images and prediction probabilities
- Generates a confusion matrix to evaluate classification performance
- Plots training and validation loss and accuracy
This project was adapted from a Coursera course Basic Image Classification with TensorFlow by Amit Yadav.