Thank you for this talk, I’m from the robotics field and had only known about the Lucas-Kanade & Farneback optical flows on OpenCV which are from way back, it’s really exciting how advanced optical flow algorithms have become!
I imagined optical flow to be such a dull ML problem personally where you regress using a dataset or a reconstruction loss. but this lecture changes my perception a bit and that there are deep theoretical issues to deal with before ML/optimization comes into the picture!
Thank you Michael, nice presentation as always! Is there a chance slides will be available again? MLSS 2013 only has video link available? Or anyone from the comments can share short notes on this topic?
Thank you for this talk, I’m from the robotics field and had only known about the Lucas-Kanade & Farneback optical flows on OpenCV which are from way back, it’s really exciting how advanced optical flow algorithms have become!
Amazing talk!!! Such great explanations
Starts getting into optical flow around 2:24.
Amazing 👌
I imagined optical flow to be such a dull ML problem personally where you regress using a dataset or a reconstruction loss. but this lecture changes my perception a bit and that there are deep theoretical issues to deal with before ML/optimization comes into the picture!
Thanks interesting application to optical flow . I appreciate .
great talk
Thank you Michael, nice presentation as always! Is there a chance slides will be available again? MLSS 2013 only has video link available? Or anyone from the comments can share short notes on this topic?
We were struck up to calculate turbulence in penstock flows. Hope we will solve the problem with OF technology
01:07:49
35:00 Blender3D never disappoints you.