Andrew Davison    @AjdDavison    11/25/2021      

From a single depth view of a pile of objects we can predict the shape and pose of all the convex objects in it, with plausible guesses even for those which are partially or fully occluded. This then allows general robot planning. Dyson Robotics Lab #ICCV2021
    8         75




AK    @ak92501    12/3/2021      

Dimensions of Motion: Learning to Predict a Subspace of Optical Flow from a Single Image abs:
    1         2

Alfredo Canziani    @alfcnz    11/25/2021      

SAVi, Slot Attention for Video, learns object-centric representations by predicting the optical flow of objects it's been cued on. This additional weak supervision signal lets the model generalise beyond the training distribution to novel objects, backgrounds, and video length.

MIT CSAIL    @MIT_CSAIL    11/5/2021      

Robotic hand manipulates thousands of objects with ease:
    61         283

AK    @ak92501    11/30/2021      

Motion-from-Blur: 3D Shape and Motion Estimation of Motion-blurred Objects in Videos abs: