This project is a challenge proposed in the course "Treinamento de Redes Neurais com Transfer Learning" of the Bootcamp "Geração Tech Unimed-BH - Ciência de Dados" of the DIO (Digital Innovation One) platform.
See the project in Google Colab here.
All the material developed during the classes can be found here.
Use Google Colab to run.
The project consists of applying the Transfer Learning method to a Deep Learning network in the Python language in the COLAB environment.
As an example, we will use the following project that performs Transfer Learning with the [MNIST Dataset]: https://colab.research.google.com/github/kylemath/ml4a-guides/blob/master/notebooks/transfer-learning.ipynb
The dataset used contains two classes: cats and dogs. A description of the database can be seen at this link. The downloadable dataset can be accessed through this other link
The project was developed as oriented and using the material made available as well as research on the web. Some things have been updated, modified or added to adapt to the use case. At the end of the project, after training the neural network, a function was created to make individual image predictions.
Henrique SK