Thomas Brox

Postdoctoral Researcher

Computer Vision Group
Department of Computer Science
U.C. Berkeley

Office location:
Sutardja Dai Hall, 7th floor
Berkeley, CA, 94720-1758

E-Mail: Brox

Thomas Brox

Personal Publications Curriculum Vitae Collaborators Demo Videos Useful Links

Videos on large displacement optical flow

Tennis sequence
Low quality video (1MB)  High quality video (36MB)  Slow motion (36MB)   Input images (173MB)

Shot from Miss Marple: A pocket full of rye
Low quality video (200kB)  High quality video (4MB)  Slow motion (4MB)

Monkey sequence
Low quality video (1.6MB)  High quality video (last part only, 12MB)  Slow motion (last part only, 12MB)

Another shot from Miss Marple: A pocket full of rye
Low quality video (500kB)  High quality video (13MB)  Slow motion (13MB)

Tennis sequence   The color code for interpreting the direction of the flow vectors is shown on the left. For instance, red means a pixel is moving to the right. Videos are shown with the original resolution and frame rate. Note that the flow fields show quantization artifacts from compression even in the high quality material. The uncompressed result is as smooth as shown in the still image.

T. Brox, C. Bregler, J. Malik
Large displacement optical flow,
IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), Miami, FL, June 2009.