Pointing at Parts: Training-Free Few-Shot Grounding in Multimodal LLMs

Part-level pointing is important for fine-grained interaction and reasoning, yet existing Multimodal Large Language Models (MLLMs) remain limited to instance-level pointing. Part-level pointing is inherently more challenging than instance-level pointing, due to the fine-grained and ambiguous nature of object parts. Instead of relying on post-training, we explore a few-shot mechanism to enhance the pointing ability of MLLMs. We introduce POinting at Parts (POP), a training-free, plug-and-play approach that addresses the challenges of part-level pointing. POP fuses textual and visual attention maps with self-supervised visual correspondences from query image and few-shot examples. On average across the three evaluated datasets, POP achieves accuracy gains of up to 8.9 points in the one-shot setting and 16.4 points in the three-shot setting for the pointing-capable MLLMs-Qwen2.5-VL, Ovis2.5, and Molmo. Notably, even MLLMs without pointing capability benefit significantly from the proposed approach. These results establish a simple yet effective path toward fine-grained spatial grounding in MLLMs.