Leeds Sports Pose is a sports pose dataset that is divided into several categories: athletics, badminton, baseball, gymnastics, parkour, football, volleyball and tennis. It contains about 2,000 pose annotations, and the images are all from athletes on Flickr.

The images are scaled so that the most prominent person is about 150 pixels long, and each image is annotated with 14 joint locations, with the left and right joints always labeled centered on the person. Additionally, properties and a Flickr link to the original image can be found in the JPEG comment field of each image file.

This dataset was released by the School of Computer Science at the University of Leeds in 2010. The related paper is "Clustered Pose and Nonlinear Appearance Models for Human Pose Estimation".