Glance2Gaze: Efficient Vision-Language Models from Glance Fusion to Gaze Compression

Open in new window