DeepPerception: Advancing R1-like Cognitive Visual Perception in MLLMs for Knowledge-Intensive Visual Grounding Paper • 2503.12797 • Published Mar 17 • 30
