Beyond 3D VQAs: Injecting 3D Spatial Priors into Vision-Language Models for Enhanced Geometric Reasoning

• URL: http://arxiv.org/abs/2605.30231v1
• Date: Thu, 28 May 2026 17:00:52 GMT
• Title: Beyond 3D VQAs: Injecting 3D Spatial Priors into Vision-Language Models for Enhanced Geometric Reasoning
• Abstract summary: Vision-Language Models (VLMs) often struggle with robust 3D spatial reasoning.<n>We propose GASP (GeometricAware Spatial Priors), a framework that injects fundamental geometric priors directly into the LLM's transformer layers.<n>GASP employs a small correspondence head, applied as a deep supervision signal across all layers, and is trained with a dual objective leveraging ground-scale video scenes.
• Score: 24.251128353444713
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Vision-Language Models (VLMs) often struggle with robust 3D spatial reasoning. Prevailing methods that rely on fine-tuning with 3D visual question-answering (VQA) datasets may overfit dataset-specific biases, while integrating specialized 3D visual encoders is often inflexible and cumbersome. In this paper, we argue that genuine spatial understanding should emerge from learning fundamental geometric priors, not only from high-level VQA supervision. We propose GASP (Geometric-Aware Spatial Priors), a framework that injects these priors directly into the LLM's transformer layers. GASP employs a small correspondence head, applied as a deep supervision signal across all layers, and is trained with a dual objective leveraging ground-truth geometry from large-scale video scenes: a contrastive loss on ground-truth point correspondences enforces 2D view-invariance, while a depth consistency supervision resolves 3D geometric ambiguities. Our analysis first provides a diagnostic showing that standard VLMs' internal correspondence matching accuracy is very low (often below 5%). We then demonstrate that our training substantially improves this behavior, boosting peak layer-wise correspondence to over 70% and maintaining over 85% temporal robustness while baselines remain below 5%. These internal improvements translate to significant gains on downstream spatial benchmarks including +18.2% on All-Angles Bench and +29.0% on VSI-Bench, all without training on any 3D VQA data. Our findings indicate that learning from fundamental geometric priors is a promising and generalizable pathway towards VLMs with more reliable 3D spatial reasoning.

Related papers

• Lift3D-VLA: Lifting VLA Models to 3D Geometry and Dynamics-Aware Manipulation [76.16065370700488]
  We introduce Lift3D-VLA, a unified VLA framework that equips models with explicit 3D point cloud reasoning and enables temporally coherent action generation.<n>We show that Lift3D-VLA achieves 10.8% and 11.1% higher mean success rates on MetaWorld and RLBench than the best-performing prior VLA methods.
  arXiv  Detail & Related papers  (2026-07-07T17:59:47Z)
• SpatialFly: Geometry-Guided Representation Alignment for UAV Vision-and-Language Navigation in Urban Environments [49.966170814478915]
  We propose a geometry-guided spatial representation framework for UAV VLN.<n>Running on RGB observations without explicit 3D reconstruction, SpatialFly introduces a geometry-guided 2D representation alignment mechanism.<n> Experimental results show that SpatialFly consistently outperforms state-of-the-art UAV VLN baselines across both seen and unseen environments.
  arXiv  Detail & Related papers  (2026-03-22T03:56:58Z)
• Learning to Reason in 4D: Dynamic Spatial Understanding for Vision Language Models [79.18306680174011]
  DSR Suite bridges gap across aspects of dataset, benchmark and model.<n>We propose an automated pipeline that generates multiple-choice question-answer pairs from in-the-wild videos for DSR.<n>The pipeline extracts rich geometric and motion information, including camera poses, local point clouds, object masks, orientations, and 3D trajectories.
  arXiv  Detail & Related papers  (2025-12-23T17:56:36Z)
• GLaD: Geometric Latent Distillation for Vision-Language-Action Models [106.53332923530245]
  We introduce GLaD, a geometry-aware Vision-Language-Action (VLA) framework that incorporates 3D geometric priors during pretraining through knowledge distillation.<n>GLaD achieves 94.1% average success rate across four LIBERO task suites, outperforming UniVLA (92.5%) which uses identical pretraining data.
  arXiv  Detail & Related papers  (2025-12-10T13:07:27Z)
• Let Language Constrain Geometry: Vision-Language Models as Semantic and Spatial Critics for 3D Generation [34.44214123004662]
  We propose VLM3D, a framework for differentiable semantic and spatial critics.<n>Our core contribution is a dual-language critic signal derived from the VLM's Yes or No log-odds.<n>VLM3D establishes a principled and general path to inject the VLM's rich, language-grounded understanding of both semantics and space into diverse 3D generative pipelines.
  arXiv  Detail & Related papers  (2025-11-18T09:05:26Z)
• Abstract 3D Perception for Spatial Intelligence in Vision-Language Models [100.13033631690114]
  Vision-language models (VLMs) struggle with 3D-related tasks such as spatial cognition and physical understanding.<n>We introduce SandboxVLM, a framework that leverages abstract bounding boxes to encode geometric structure and physical kinematics for VLM.<n>Our approach consistently improves spatial intelligence, achieving an 8.3% gain on SAT Real compared with baseline methods.
  arXiv  Detail & Related papers  (2025-11-14T04:16:09Z)
• 3D-Aware Vision-Language Models Fine-Tuning with Geometric Distillation [17.294440057314812]
  Vision-Language Models (VLMs) have shown remarkable performance on diverse visual and linguistic tasks.<n>We propose Geometric Distillation, a framework that injects human-inspired geometric cues into pretrained VLMs.<n>Our method shapes representations to be geometry-aware while remaining compatible with natural image-text inputs.
  arXiv  Detail & Related papers  (2025-06-11T15:56:59Z)
• Virtual Normal: Enforcing Geometric Constraints for Accurate and Robust Depth Prediction [87.08227378010874]
  We show the importance of the high-order 3D geometric constraints for depth prediction. By designing a loss term that enforces a simple geometric constraint, we significantly improve the accuracy and robustness of monocular depth estimation. We show state-of-the-art results of learning metric depth on NYU Depth-V2 and KITTI.
  arXiv  Detail & Related papers  (2021-03-07T00:08:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.