arXiv Computer Vision

Multi-View Relational Distillation for Spatial Reasoning with Vision-Language Models

The paper introduces Multi-View Relational Distillation (MVRD), a method that distills patch-wise cosine similarities across multiple views from geometry‑grounded vision models into vision‑language models (VLMs). By focusing on relational geometry rather than raw teacher features, MVRD preserves the pretrained alignment between visual and textual representations while enhancing spatial reasoning. Experiments show that MVRD outperforms supervised fine‑tuning and feature distillation, approaching feature‑fusion performance with fewer parameters and lower latency, and it generalizes to 3D scene understanding tasks such as object grounding, dense captioning, and question answering.

arXiv Computer Vision
Sep 30

Spatial-OPSD: Self-Improving Spatial Reasoning via Label-Free Self-Distillation

Spatial-OPSD is a label‑free self‑improvement framework for vision‑language models that leverages spatial priors such as depth, 3D relations, and camera geometry to provide dense token‑level supervision. During training, a privileged teacher uses these priors while the student learns from only the original visual‑language input, and a recursive round‑wise scheme allows repeated self‑improvement without moving the teacher. Across four VLM families, one round of Spatial‑OPSD improves the five‑benchmark average, and three rounds push a strong spatially specialized model to the open‑source frontier, achieving the highest average among open models and best results on three of five spatial reasoning benchmarks.

By Zhenyu Liu, Zhangquan Chen, Keyi Chen, Mingze Sun, Xiang An, Haodong Jing, Ruqi Huang
arXiv AI
Sep 10

MV-STRIDE: Enabling MLLMs to Master Multi-View Spatial Reasoning via Hierarchical Capability Modeling

MV-STRIDE is a Multi‑View hierarchical Spatial Reasoning dataset that models dependencies among perception, scene understanding, and contextual reasoning to support 3D spatial cognition. It introduces a QA generation pipeline that enforces cross‑view constraints, producing multi‑level reasoning tasks with chain‑of‑thought supervision. Experiments show that training on MV‑STRIDE yields state‑of‑the‑art performance on multi‑view spatial benchmarks, enabling MLLMs to reason robustly across diverse viewpoints.

By Jin Xu, Xiaojian Huang, Zhuodong Luo, Zhihong Zhang, Xin Liu, Jiansheng Wei, Xinzhi Wang, Jie Zhao, Xuejin Chen
arXiv Computation and Language
Sep 30

Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering

arXiv:2609.38177v1 Announce Type: cross Abstract: Reasoning about the 3D world from multi-view images remains a fundamental challenge for Multimodal Large Language Models (MLLMs). While modern MLLMs...

By Jaewoo Jung, Hyeonseo Yu, Honggyu An, Jisang Han, Mungyeom Kim, Minkyeong Jeon, Heeseong Shin, Wonjun Moon, Federico Tombari, Daniel Barath, Marc Pollefeys, Seungryong Kim, Sunghwan Hong
arXiv Computer Vision
Oct 9

Seek-and-View Reasoning for Multi-View Spatial Understanding

The paper introduces Vantage, a training‑free, model‑agnostic framework that pairs a vision‑language model with a 3D foundation model to perform Seek‑and‑View reasoning for multi‑view spatial understanding. Vantage first analyzes a question to plan a relevant viewpoint, then augments reasoning with geometry‑grounded visual evidence from that viewpoint. Experiments across six VLMs on five benchmarks show consistent improvements without fine‑tuning, demonstrating that revealing spatial evidence through view‑grounded reasoning reduces reliance on language‑based cross‑view alignment.

By Qixiang Chen, Cheng Zhang, Fucai Ke, Chi-Wing Fu, Jianfei Cai, Jingwen Ye
arXiv Computer Vision
Oct 16

DK-VLM: Geometry as Position for Spatial Reasoning in Vision-Language Models

DK‑VLM introduces Dual‑Kernel Attention, where a visual token’s attention is computed from both its input position and its geometric position derived from 3D points and camera poses. A semantic kernel preserves pretrained attention while a geometric kernel modulates attention via point displacements and relative view transformations, with a learned gate fusing the two. The model also employs Reference‑Frame Composition supervision to update its reference frame across successive judgments, achieving top scores on six spatial benchmarks and strong performance on VSI‑Bench and ReVSI‑32.

By Han Su, Meng Tian, Tianyu Huang, Mingyu Gao, Lennox Wei, Dongxu Li, Jianhua Han, Hang Xu, Xiaohe Wu, Wangmeng Zuo
arXiv AI
Oct 9

Distilling Routed 3D Privilege for Spatial Reasoning in Vision-Language Models

The paper introduces Geometry-Privileged Distillation (GPD), a method that incorporates depth, semantic, and bird’s-eye-view cues as compact text routed to a teacher model during on-policy self-distillation. By applying a privileged KL loss only to incorrect trajectories, GPD enhances spatial reasoning in vision‑language models while keeping the deployed model RGB‑only. Experiments on a 4B backbone show that GPD outperforms prior methods on multiple spatial reasoning benchmarks, with ablations confirming the benefits of 3D cues, question‑conditioned routing, and selective distillation.

By Hongxing Li, Yixin Li, Dingming Li, Zixuan Wang, Yuchen Yan, Wenqi Zhang, Weiming Lu, Yongliang Shen
arXiv AI
Jul 21

Spatiotemporal Knowledge Graphs as Persistent Scene Memory for Embodied Question Answering

arXiv:2510. 01483v3 Announce Type: replace-cross Abstract: Vision-language models (VLMs) demonstrate strong image-level scene understanding, but reasoning over long egocentric video remains costly: because VLMs maintain no persistent memory or explicit spatial representation, all sampled frames must be re-processed for every new query.

By Mohamad Al Mdfaa, Svetlana Lukina, Timur Akhtyamov, Arthur Nigmatzyanov, Dmitrii Nalberskii, Sergey Zagoruyko, Gonzalo Ferrer