Tiny Images Dataset is a tiny image dataset containing 79,300,000 32*32 pixel color images collected from Google Images and divided into 75,000 categories. The Tiny Images Dataset is often used in conjunction with the ImageNet training set to benchmark computer vision algorithms. This dataset was created by MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) in 2006 to promote the development of target detection technology. Related papers include "80 million tiny images: a large dataset for non-parametric object and scene recognition".