Download the CIFAR-10 dataset and unzip. Clone repository in unzipped folder and install dependencies with npm install. Run parser for example with node parser.js -i ./../data_batch_1.bin -p
This initial code is a way to get to grips with the CIFAR-10 dataset & prepare for usage in other systems like Torch.
Currently writing images to disk is blocking as the read loop waits for a callback from writing an RGB buffer to png. Can be paralleled with promises or other throttling.
Krizhevsky, A., 2009. Learning Multiple Layers of Features from Tiny Images. , p.1-60.