PreLAR: World Model Pre-training with Learnable Action Representation.

Lixuan Zhang, Meina Kan, Shiguang Shan, Xilin Chen

Browse the full ECCV paper archive.