Datasets

Before you can do any sort of machine learning task, you need data. Fortunately there are a great variety of datasets out there for all sorts of things.

Here are some websites which have large repositories of datasets:

• Dataset List
• Deep Learning datasets
• Google Dataset Search
• Kaggle datasets
• UCI Machine Learning Repository

Some basic image datasets:

• CIFAR-10
• CIFAR 100
• Fashion MNIST
• Caltech 101

Face recognition datasets:

• Face Recognition Datasets
• Labeled Faces in the Wild

Object detection datasets:

• German Traffic Sign Classification
• KITTI Dataset
• Street View House Numbers (SVHN)

Downloading Datasets from Kaggle

You can download datasets from Kaggle through the command line using the Kaggle API. You can install the tool through pip in your Anaconda environment. You’ll also have to create an account on Kaggle and generate API credentials (follow the instructions in the README). When you visit a dataset the Kaggle website, it will provide a command that you can copy and paste to download that dataset via command line.