Abstract: Recent open-world representation learning approaches have leveraged CLIP to enable zero-shot 3D object recognition. However, performance on real point clouds with occlusions still falls ...
Abstract: We present TabGuard, a privacy-preserving framework for an end-to-end secure Table Structure Recognition. Tab-Guard masks all the contents of the table locally and utilizes the masked table ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results