Integrating Motion into Vision Models for Better Visual Prediction

Michael Hazoglou, Todd Hylton

We demonstrate an improved vision system that learns a model of its environment using a self-supervised, predictive learning method. The system includes a pan-tilt camera, a foveat…