Dataset to CIFAR

This package was created to process datasets of images (mainly for Machine Learning purposes) into the CIFAR-10 binary version format. With this package you can:

• Turn your image datasets into the binary version of the CIFAR-10 format
• Resize the images to a desire dimension before processing
• Split the image datasets into a desired number of batches.

Please refer to the official website for CIFAR-10 for more information on how the data is formatted.

Dependencies

pip install --upgrade Pillow

How to use?

Images beloging to an specific class (i.e. Trees) need to get stored into separate directories where the name of the directory is the class label.

TODO:

• Test on python 2.7x
• Transform image datasets into CIFAR-10 python version and CIFAR-10 Matlab version
• Improve running time