HorizonRobotics/ARSG-110K
收藏资源简介：
--- license: apache-2.0 task_categories: - image-to-3d tags: - 3d - scene-generation - cvpr --- # ARSG-110K [**Project Page**](https://zx-yin.github.io/3dfixer/) | [**Paper**](https://huggingface.co/papers/2604.04406) | [**GitHub**](https://github.com/HorizonRobotics/3D-Fixer) ARSG-110K is a large-scale scene-level dataset comprising over 110K diverse scenes and 3M annotated images with high-fidelity 3D ground truth. It is designed to support the training and evaluation of compositional 3D scene generation and in-place completion models. The dataset provides accurate 3D object-level ground-truth, layout, and annotations. This dataset was introduced as part of the paper: **3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes from a Single Image**. ## Dataset Details - **Source Data**: The `Objaverse_github.csv` and `ObjaverseXL_sketchfab.csv` are subsets of those from [TRELLIS-500K](https://github.com/microsoft/TRELLIS?tab=readme-ov-file#-dataset). - **Assets**: The `hdrs.zip`, `materials_floor.zip`, and `materials_wall.zip` contain assets from [BlenderKit](https://www.blenderkit.com/). - **Examples**: This repository provides renderings of 1000 example scenes. ## Accessing the Data Please follow the [detailed instructions](https://github.com/HorizonRobotics/3D-Fixer/blob/main/DATASET.md) on the official GitHub repository to access the full dataset beyond the example renderings. ## Citation ```bibtex @inproceedings{yin2026tdfixer, title={3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes from a Single Image}, author={Yin, Ze-Xin and Liu, Liu and Wang, Xinjie and Sui, Wei and Su, Zhizhong and Yang, Jian and Xie, jin}, booktitle={Proceedings of the Computer Vision and Pattern Recognition Conference}, year={2026} } ```
许可证：Apache-2.0 任务类别： - 图像转3D（image-to-3d） 标签： - 3D - 场景生成 - 计算机视觉与模式识别会议（CVPR） # ARSG-110K [**项目主页**](https://zx-yin.github.io/3dfixer/) | [**论文**](https://huggingface.co/papers/2604.04406) | [**GitHub仓库**](https://github.com/HorizonRobotics/3D-Fixer) ARSG-110K是一款大规模场景级数据集，涵盖超11万个多样化场景与300万张带有高保真3D基准真值（ground truth）的标注图像。其研发目标为支撑组合式3D场景生成与原位补全模型的训练与评估工作。该数据集提供精确的3D物体级基准真值、场景布局及各类注释信息。 本数据集随论文**《3D-Fixer：基于单图像的3D场景粗到细原位补全》**一同发布。 ## 数据集详情 - **源数据**：`Objaverse_github.csv`与`ObjaverseXL_sketchfab.csv`为[TRELLIS-500K](https://github.com/microsoft/TRELLIS?tab=readme-ov-file#-dataset)数据集的子集。 - **素材资源**：`hdrs.zip`、`materials_floor.zip`与`materials_wall.zip`包含来自[BlenderKit](https://www.blenderkit.com/)的素材资源。 - **示例场景**：本仓库提供1000个示例场景的渲染结果。 ## 数据集获取方式 请参照官方GitHub仓库中的[详细指南](https://github.com/HorizonRobotics/3D-Fixer/blob/main/DATASET.md)，获取示例渲染结果之外的完整数据集。 ## 引用格式 bibtex @inproceedings{yin2026tdfixer, title={3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes from a Single Image}, author={Yin, Ze-Xin and Liu, Liu and Wang, Xinjie and Sui, Wei and Su, Zhizhong and Yang, Jian and Xie, jin}, booktitle={Proceedings of the Computer Vision and Pattern Recognition Conference}, year={2026} }
