license: apache-2.0
task_categories:
- video-text-to-text
tags:
- 3D
- vision-language
- spatial-intelligence
SpaceSpan Dataset
SpaceSpan is a large-scale dataset curated for aligning 3D proxy representations with Vision-Language Models (VLMs), introduced in the paper Proxy3D: Efficient 3D Representations for Vision-Language Models via Semantic Clustering and Alignment.
The dataset incorporates heterogeneous visual information into a unified format to support multi-stage training for developing spatial intelligence. It enables models to progress from simple image-text alignment to complex 3D reasoning tasks, such as 3D visual question answering (VQA) and visual grounding.
Project Page | GitHub | Paper
Dataset Description
The SpaceSpan dataset (specifically the SpaceSpan-318K version) supports four progressive training stages:
- Stage 1: Initial spatial alignment.
- Stage 2-3: Intermediate spatial reasoning development.
- Stage 4: Full-scale 3D reasoning.
Directory Structure
Based on the official repository, the dataset is typically organized as follows:
data/ # Training and inference data
βββ icon_image_embeds_qwen25.pt
βββ number_image_embeds_qwen25.pt
βββ stage_1_train.json
βββ stage_2_train.json
βββ stage_3_train.json
βββ stage_4_train_318K.json
βββ pointmaps_wo_markers
βββ poses
βββ ...
Citation
If you find this dataset useful for your research, please cite the following paper:
@article{proxy3d2026,
title={Proxy3D: Efficient 3D Representations for Vision-Language Models via Semantic Clustering and Alignment},
author={Jiang, Jerry and Sun, Haowen and Gudovskiy, Denis and Nakata, Yohei and Okuno, Tomoyuki and Keutzer, Kurt and Zheng Wenzhao},
journal={arXiv preprint arXiv:2605.08064},
year={2026}
}