[ CODE ]

Learning to Count without Annotations

Geometric Superpixel Representations for Efficient Image Classification with Graph Neural Networks

Self-Supervised Learning of Structured Dynamics from Videos