Abstract: Object-centric representation learning aims to decompose visual scenes into fixed-size vectors called "slots" or "object files", where each slot captures a distinct object. Current ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback