This is the beginning of a series of projects working with Convolutional Neural Networks (CNNs).
- Python 3
- Tensorflow >= 1.10
- Numpy
- Pandas
- Scikit-learn
- Matplotlib
I'd recommend using Anaconda as a package manager and the accompanying Spyder IDE.
I prefer Anaconda because of the ease with which you can set up environments to keep
package installations separate, but obviously you can use anything you like.
Working over a notebook is a cleaner way of approaching simpler machine learning problems. Using Google Colab makes it more straightforward to mount data over Drive and then import it for training/validationt/testing.
This project is an animal image classification problem.
I'm using this dataset consisting of 13,000 animal images across 30 categories to train/test with, and 6,000 images to predict using the trained model:
https://s3-ap-southeast-1.amazonaws.com/he-public-data/DL%23+Beginner.zip