Geometry-Based Next Frame Prediction from Monocular Video

  Abstract