Explainable and Resource-Efficient Spatial Reasoning in Multimodal LLMs for Decision-Critical Applications
2026-07-29 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors improved a tool that helps large language models understand spatial relationships in images, which is important for tasks like robotics and safety monitoring. Their new method, ByDeWay-V2, adds clear descriptions of how objects relate to each other (like "left of" or "inside") along with depth information to reduce mistakes and make the model's reasoning easier to check. They tested their approach on benchmarks and found it significantly better at understanding spatial clues than before, even when using limited computing power. This shows their method can help machines make better spatial decisions without extra training.
Multimodal Large Language Models (MLLMs)spatial reasoningobject hallucinationLayered-Depth-Based Prompting (LDP)monocular depth estimationspatial predicatesopen-vocabulary object detectionVisual Spatial Reasoning (VSR)BLINK benchmarkprompt engineering
Authors
Piyush Jain, Kousik Dasgupta, Rajarshi Roy, Subarna Tripathi
Abstract
As Multimodal Large Language Models (MLLMs) are increasingly deployed in decision-critical pipelines such as robotics, embodied AI, and safety monitoring, the opacity of their spatial judgments limits operator trust and auditability. MLLMs demonstrate strong reasoning but often struggle with fine-grained spatial understanding and object hallucination. Prior work, ByDeWay, introduced Layered-Depth-Based Prompting (LDP), a training-free framework that mitigates hallucinations by structuring prompts using monocular depth estimation. However, coarse depth layering falls short in resolving object-to-object spatial relationships within the same geometric plane, such as projective ("left of", "above") and topological ("inside", "touching") relations. We propose ByDeWay-V2, which integrates explicit spatial relational context alongside depth cues, expressed as human-readable predicates that serve as auditable evidence for downstream decision support. Using an open-vocabulary object detector (YOLO-World-L), our framework computes pairwise geometric relations between detected objects and injects them as structured spatial predicates into the MLLM prompt, bridging 3D scene depth and 2D spatial semantics without any training. We evaluate ByDeWay-V2 on the Visual Spatial Reasoning (VSR) and BLINK benchmarks across multiple MLLMs, with hallucination grounding assessed via POPE. On the BLINK spatial subset, ByDeWay-V2 achieves a 46 percent relative F1 improvement over LDP for Qwen2.5-VL, and recovers BLIP-Base's spatial reasoning on VSR from near-random performance to a competitive F1 of 0.53. Our lightest configuration operates under a strict 40-token context budget on CPU, showing the framework's suitability for resource-constrained, real-time decision-support settings.