This is a tutorial project refactored from the Udacity course: Intro to TensorFlow for Deep Learning.
The objective of this project was to create and train my first neural network that recognizes images of clothing. To focus attention away from environment and package management, I chose to develop within Google Colab. You can run my project there:
The dataset provides 70,000 low-resolution images of clothing articles within 10 categories (shirts, sandels, pants etc.). Each sample is a 24x24 pixel image, as shown below.
Fashion-MNIST samples (by Zalando, MIT License).
This neural network leverages a number of libraries:
- TensorFlow - machine learning platform.
- TensorFlow Datasets - collection of datasets ready to use in TensorFlow.
- Keras - neural network API.
- numpy - scientific computing functions.
- matplotlib - visualization tools.
MIT