![]() |
Supplementary Material Page for Our ECCV Paper Joint Estimation of Motion, Structure and |
Home |
This page shows the results for our approach applied to the longer "Pascal"
sequence. This is a real-world sequence of approximately 70 frames that shows
different facial expressions. Four videos with our results are shown here: two
for the textured model and two for the untextured model. The scene flow is color
coded, where green stands for small displacement vectors and red for large
displacement vectors. The untextured result can be interpreted as a
3-dimensional variant of optical flow. The first textured video can be
downloaded
here [6.2 MB] and is shown below:
Please enable Javascript and Flash to see the video
Please enable Javascript and Flash to see the video
Please enable Javascript and Flash to see the video
Please enable Javascript and Flash to see the video
Below we show detailed results of two expressions: Opening Eyes and
Raising Eyebrows.
![]() ![]()
Left: Left image at time t.
Right: Right image at time t+1. ![]()
Scene flow. ![]() Overlay of reconstruction and scene flow. ![]()
Zoom in previous image.
![]() ![]()
Left: Left image at time t.
Right: Right image at time t+1. ![]()
Scene flow. ![]() Overlay of reconstruction and scene flow. ![]() Zoom in previous image. |
MIA Group |