Adapting temporal action segmentation models from exocentric to egocentric scenarios using synchronized video pairs and knowledge distillation.
