PaperActive

SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem

quality59

Updated 29 min ago · first seen 11 Sept 2026

paper_01M294WYDPBZCXJ1FMAK30Z44W

Published: 7 Sept 2026; T2 · 6 h ago
arXiv: 2609.07064; T2 · 6 h ago

Abstract

Large Vision-Language Models (LVLMs) have achieved strong performance on diverse visual tasks, yet their ability to reconstruct and reason about the 3D structure of the scene depicted in 2D images -- referred to as spatial intelligence -- remains limited. Existing approaches attempt to address this gap by using real-scene spatial question answering datasets that require dense geometric annotations. However, constructing such labels is costly, time-consuming, and often noisy due to reliance on external perception modules. In this work, we propose a novel paradigm inspired by human cognitive development: learning foundational spatial skills through structured block-manipulation tasks. We introduce SpatialBlock-15k, a synthetic dataset of 15,000 block-stacking problems covering 3D-to-2D projection, viewpoint transformation, and structural combination. The dataset further incorporates controlled color modulation as visual cues to encourage anchor-based reasoning in visually complex conditions. Experiments demonstrate that LVLMs trained on our dataset through either direct answering or reasoning-based prediction significantly outperform baselines and generalize to real-world spatial tasks, despite the dataset's synthetic and compact nature. Code and data are available at https://github.com/rsoohyun/SpatialBlock.

Authors 3

Soohyun Ryu, Sohee Kim, Eunho Yang

Specification

Official page: arxiv.org/abs/2609.07064
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 6 h agomedium
arXiv id: 2609.07064
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 6 h agomedium
Github repo: rsoohyun/SpatialBlock
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 6 h agomedium
Hf paper url: https://huggingface.co/papers/2609.07064
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 6 h agomedium
Github stars: 2
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 29 min agomedium
Hf comments: 2
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 29 min agomedium
Upvotes: 68
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 29 min agomedium
PDF: arxiv.org/pdf/2609.07064
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 6 h agomedium
Published: 7 Sept 2026
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 6 h agomedium

Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →

Provenance

Attributed facts

Source tiers

T211

Freshest observation

29 min ago

Conflicts

None

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history · Github repo

1 claims · 1 propertiesShow all properties

Github repogithub_repo1

Claim history for Github repo
Value	Valid from → to	Status	Source	Confidence	Extractor
rsoohyun/SpatialBlock	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/papers	listing	T2· Quality secondary	29 min ago	6

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.