KITTI

FitVid predictions on held-out videos.

• The left image is the ground truth.

• The right image is the predicted video.

  • The red border indicates the context frames.

  • The green border means the frame was predicted.