STL-10 Dataset is an image recognition dataset used to develop unsupervised feature learning, deep learning, and self-learning algorithms. It is based on CIFAR-10 Dataset with some modifications. Each category contains fewer labeled training examples than CIFAR-10, but provides larger unlabeled instances for supervised training of image learning models, which use unlabeled data to build prior data, and high-resolution images can be used to develop more scalable unsupervised learning methods.

This dataset was released by Stanford University in 2011. The main publishers were Adam Coates, Honglak Lee and Andrew Y. Ng. The related paper is "An Analysis of Single Layer Networks in Unsupervised Feature Learning".