The Stanford Drone Dataset is a large dataset consisting of images and videos of various types of pedestrians, covering eight scenarios in crowded spaces, such as a university campus or the sidewalk of a busy street.

Humans follow certain guidelines when walking in environments such as university campuses and sidewalks. This dataset aims to design a new set of algorithms to solve tasks such as target tracking and trajectory prediction.

Stanford Drone Dataset was released by the Computer Vision and Geometry Laboratory of Stanford University in 2016. The related paper is "Learning Social Etiquette: Human Trajectory Prediction In Crowded Scenes".