See&Trek: Training-Free Spatial Prompting for Multimodal Large Language Model
This addresses the underexplored challenge of purely visual-spatial reasoning for MLLM users, offering an incremental improvement through a novel prompting method.
The paper tackles the problem of enhancing spatial understanding in Multimodal Large Language Models under vision-only constraints by introducing SEE&TREK, a training-free prompting framework that improves performance on spatial reasoning tasks by up to +3.5% on benchmarks like VSI-BENCH and STI-BENCH.
We introduce SEE&TREK, the first training-free prompting framework tailored to enhance the spatial understanding of Multimodal Large Language Models (MLLMS) under vision-only constraints. While prior efforts have incorporated modalities like depth or point clouds to improve spatial reasoning, purely visualspatial understanding remains underexplored. SEE&TREK addresses this gap by focusing on two core principles: increasing visual diversity and motion reconstruction. For visual diversity, we conduct Maximum Semantic Richness Sampling, which employs an off-the-shell perception model to extract semantically rich keyframes that capture scene structure. For motion reconstruction, we simulate visual trajectories and encode relative spatial positions into keyframes to preserve both spatial relations and temporal coherence. Our method is training&GPU-free, requiring only a single forward pass, and can be seamlessly integrated into existing MLLM'S. Extensive experiments on the VSI-B ENCH and STI-B ENCH show that S EE &T REK consistently boosts various MLLM S performance across diverse spatial reasoning tasks with the most +3.5% improvement, offering a promising path toward stronger spatial intelligence.