AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Attention-based Vision-Language Memory for Spatial Reasoning

Zuntao Liu, Zuntao Liu, Taimeng Fu, Shaoshu Su, Cherie Ho, Chen Wang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add 3D-aware representation (ablation)LLaVA-NeXT-Video-7B baseline · VSI-Bench avgaccuracy: 55.2 -> 63.8Improved
add working memory (ablation)VLM2 3D-aware · VSI-Bench avgaccuracy: 63.8 -> 65.9Improved
add episodic memory (ablation)VLM2 3D-aware · VSI-Bench avgaccuracy: 63.8 -> 66.1Improved
add both memory modules (ablation)VLM2 3D-aware · VSI-Bench avgaccuracy: 63.8 -> 67.8Improved
use cross-attention instead of concat-MLP fusion (ablation)pi3 geometry · VSI-Bench avgaccuracy: 40.4 -> 61.0Improved
add adaptive gate to 3D position injection (ablation)A3PI · VSI-Bench avgaccuracy: 58.9 -> 61.6Improved
add viewpoint-aware geometry alignment (ablation)Cross-Attn pi3 baseline · VSI-Bench avgaccuracy: 61.0 -> 62.9Improved
combine A3PI and VAGA (ablation)Cross-Attn pi3 baseline · VSI-Bench avgaccuracy: 61.0 -> 63.8Improved
use pi3 instead of VGGT geometry with cross-attention (ablation)Cross-Attn fusion · VSI-Bench avgaccuracy: 60.2 -> 61.0Improved
increase episodic memory capacity from 8 to 32 (ablation)VLM2 (Lw=8) · VSI-Bench avgaccuracy: 66.3 -> 68.8Improved
increase working memory window from 8 to 16 (ablation)VLM2 (Le=32) · VSI-Bench avgaccuracy: 68.8 -> 68.3Hurt
use VLM2 instead of VLM-3R on long videosVSI-Bench · VSI-Bench longaccuracy: 60.0 -> 68.1Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy