CIFAR-10 image classifier using TensorFlow that demonstrates:
- Download of CIFAR-10 dataset using Keras
- Scaling of image data and one-hot encoding of the labels
- Defining a simple CNN using the tf.layers API
- Saving data and graph for display with TensorBoard
- Saving checkpoints and graph as protobuf files
- Creation of frozen graph for deployment
The Jupyter notebook can be visualized or downloaded from here.
The Python source code is here.