Paper 2603.07660
Holi-Spatial: Evolving Video Streams into Holistic 3D Spatial Intelligence
- Published
- Mar 2026
- Research lab
- Independent
- Citations
- 5
- GitHub
- 373 stars
01 In brief
Summary
Holi-Spatial is a fully automated pipeline that converts raw video streams into holistic 3D spatial annotations without human intervention, addressing the scalability limits of existing spatial intelligence datasets that rely on manually annotated 3D scans.
The pipeline has three stages: geometric optimization using 3D Gaussian Splatting (3DGS) with monocular depth priors, image-level perception using a VLM and SAM3 for open-vocabulary segmentation, and scene-level refinement that merges, filters, and verifies 3D proposals.
The authors release Holi-Spatial-4M, a large-scale dataset with 12K optimized 3DGS scenes, 1.3M 2D masks, 320K 3D bounding boxes, 320K instance captions, 1.2M 3D grounding instances, and 1.2M spatial QA pairs.
Evaluations on ScanNet, ScanNet++, and DL3DV show Holi-Spatial outperforms baselines in depth estimation, 2D segmentation, and 3D detection, e.g., improving depth F1 by 0.5 and 3D detection AP50 by 64% on ScanNet++.
Fine-tuning Qwen3-VL on Holi-Spatial-4M yields a 15% AP50 gain on ScanNet++ 3D grounding and a 7.9% accuracy increase on MMSI-Bench.
The pipeline is scalable as resources permit, though it relies on multiple upstream components and per-scene optimization, which can be computationally expensive and may degrade under challenging videos.
02 From the paper
Abstract
The pursuit of spatial intelligence fundamentally relies on access to large-scale, fine-grained 3D data. However, existing approaches predominantly construct spatial understanding benchmarks by generating question-answer (QA) pairs from a limited number of manually annotated datasets, rather than systematically annotating new large-scale 3D scenes from raw web data. As a result, their scalability is severely constrained, and model performance is further hindered by domain gaps inherent in these narrowly curated datasets. In this work, we propose Holi-Spatial, the first fully automated, large-scale, spatially-aware multimodal dataset, constructed from raw video inputs without human intervention, using the proposed data curation pipeline. Holi-Spatial supports multi-level spatial supervision, ranging from geometrically accurate 3D Gaussian Splatting (3DGS) reconstructions with rendered depth maps to object-level and relational semantic annotations, together with corresponding spatial Question-Answer (QA) pairs. Following a principled and systematic pipeline, we further construct Holi-Spatial-4M, the first large-scale, high-quality 3D semantic dataset, containing 12K optimized 3DGS scenes, 1.3M 2D masks, 320K 3D bounding boxes, 320K instance captions, 1.2M 3D grounding instances, and 1.2M spatial QA pairs spanning diverse geometric, relational, and semantic reasoning tasks. Holi-Spatial demonstrates exceptional performance in data curation quality, significantly outperforming existing feed-forward and per-scene optimized methods on datasets such as ScanNet, ScanNet++, and DL3DV. Furthermore, fine-tuning Vision-Language Models (VLMs) on spatial reasoning tasks using this dataset has also led to substantial improvements in model performance.