People detection from a UAV

For pedestrian detection in outdoor surveillance scenarios, the size of pedestrians in the images are often very small (around 20 pixels tall). The most common and successful approaches for single-frame pedestrian detection use gradient-based features and a support vector machine classifier. Colorado School of Mines has developed a new algorithm that extracts gradient features from a spatio-temporal volume, consisting of a short sequence of images (about one second in duration). The additional information provided by the motion of the person compensates for the loss of resolution.