Paper

Reinforcing Egocentric Spatial Perception in Multimodal Large Language Models via Ego Scene Augmentation

arXiv:2607.14497v1 Announce Type: new Abstract: Egocentric Visual Question Answering (VQA) has attracted widespread attention as an important task for enabling Multimodal Large Language Models (MLLMs) to interact with the real world. However, existing MLLMs struggle to perform effective spatial reasoning in complex egocentric scenes due to their limited spatial perception capabilities. To this end, we introduce Ego Scene Augmentation (ESA), an egocentric spatial perception framework, which actively enhances the spatial perception capabilities from the egocentric perspective, powered by the pr…

Authors:

Related discussions