A simple project about how to train a neural network with very few samples (about 300 images per class and 4 classes) using Transfer Learning in Keras & Tensorflow.
In this project I use a pre trained Neural Network as a feature extractor and them use those features to train a Logistic Regression model for identifying between 4 classes (cups, keyboards, backpack bags and bottles).
- Tensorflow 2.1.0
- Keras 2.3.1
- scikit-learn 0.22.1