Datasets:
Formats:
parquet
Languages:
English
Size:
10K - 100K
Tags:
rgb-d
indoor
scene-understanding
semantic-segmentation
3d-object-detection
scene-classification
License:
Add dataset card and animated banner
#2
by dronefreak - opened
- README.md +155 -0
- assets/demo_banner.mp4 +3 -0
- assets/demo_banner_poster.jpg +3 -0
README.md
CHANGED
|
@@ -1,4 +1,24 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
dataset_info:
|
| 3 |
features:
|
| 4 |
- name: image
|
|
@@ -66,3 +86,138 @@ configs:
|
|
| 66 |
- split: test
|
| 67 |
path: data/test-*
|
| 68 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
license: other
|
| 5 |
+
size_categories:
|
| 6 |
+
- 10K<n<100K
|
| 7 |
+
task_categories:
|
| 8 |
+
- image-segmentation
|
| 9 |
+
- depth-estimation
|
| 10 |
+
- object-detection
|
| 11 |
+
pretty_name: SUN RGB-D
|
| 12 |
+
license_name: sun-rgbd-no-license-specified
|
| 13 |
+
license_link: https://rgbd.cs.princeton.edu/
|
| 14 |
+
tags:
|
| 15 |
+
- rgb-d
|
| 16 |
+
- indoor
|
| 17 |
+
- scene-understanding
|
| 18 |
+
- semantic-segmentation
|
| 19 |
+
- 3d-object-detection
|
| 20 |
+
- scene-classification
|
| 21 |
+
- computer-vision
|
| 22 |
dataset_info:
|
| 23 |
features:
|
| 24 |
- name: image
|
|
|
|
| 86 |
- split: test
|
| 87 |
path: data/test-*
|
| 88 |
---
|
| 89 |
+
|
| 90 |
+
<h1 align="center">SUN RGB-D: Indoor RGB-D Scene Understanding Benchmark</h1>
|
| 91 |
+
|
| 92 |
+
<p align="center"><video controls autoplay loop muted playsinline width="900" poster="https://huggingface.co/datasets/dronefreak/SUN-RGBD/resolve/main/assets/demo_banner_poster.jpg" src="https://huggingface.co/datasets/dronefreak/SUN-RGBD/resolve/main/assets/demo_banner.mp4"><img src="https://huggingface.co/datasets/dronefreak/SUN-RGBD/resolve/main/assets/demo_banner_poster.jpg" alt="SUN RGB-D test frames from six scene types cycling through RGB, depth and 37-class segmentation" width="900"></video></p>
|
| 93 |
+
|
| 94 |
+

|
| 95 |
+

|
| 96 |
+

|
| 97 |
+

|
| 98 |
+

|
| 99 |
+

|
| 100 |
+
|
| 101 |
+
> **Unofficial redistribution of SUN RGB-D in a viewer-ready Parquet layout: RGB, raw and filled depth, 37- and 13-class semantic labels, scene category, camera parameters, and 2D/3D object annotations in one row per frame.**
|
| 102 |
+
|
| 103 |
+
## Dataset Description
|
| 104 |
+
|
| 105 |
+
- **Homepage:** https://rgbd.cs.princeton.edu/
|
| 106 |
+
- **Paper:** https://rgbd.cs.princeton.edu/paper.pdf
|
| 107 |
+
- **Label source:** https://github.com/ankurhanda/sunrgbd-meta-data
|
| 108 |
+
- **License:** Not specified by the authors
|
| 109 |
+
|
| 110 |
+
## Disclaimer
|
| 111 |
+
|
| 112 |
+
This repository is **not** an official release of SUN RGB-D.
|
| 113 |
+
|
| 114 |
+
SUN RGB-D was created by Shuran Song, Samuel P. Lichtenberg and Jianxiong Xiao (Princeton University). It includes images from NYU Depth v2 (Silberman et al.), Berkeley B3DO (Janoch et al.) and SUN3D (Xiao et al.). The original authors retain all copyright and intellectual property rights; this repository does **not** claim ownership of any images, annotations, or metadata.
|
| 115 |
+
|
| 116 |
+
**No license was specified by the original authors.** The project page only asks users to cite the SUN RGB-D paper and the three source datasets. This copy is an unofficial redistribution for research purposes, with full attribution. **If you are an author or rights holder and want this repository removed or changed, please open a discussion on this repository and it will be addressed promptly.**
|
| 117 |
+
|
| 118 |
+
The 37- and 13-class semantic labels are the extraction by Ankur Handa ([sunrgbd-meta-data](https://github.com/ankurhanda/sunrgbd-meta-data), no license specified), made from the official `SUNRGBD2Dseg.mat`.
|
| 119 |
+
|
| 120 |
+
---
|
| 121 |
+
|
| 122 |
+
# Dataset Overview
|
| 123 |
+
|
| 124 |
+
SUN RGB-D contains 10,335 indoor RGB-D images captured with four different sensors (Intel RealSense, Asus Xtion, Microsoft Kinect v1 and v2), densely annotated with 146,617 2D polygons and 58,657 3D bounding boxes, plus scene categories and room layouts. It is a standard benchmark for indoor semantic segmentation, scene classification, and 3D object detection.
|
| 125 |
+
|
| 126 |
+
---
|
| 127 |
+
|
| 128 |
+
# Changes from the Official Release
|
| 129 |
+
|
| 130 |
+
- Repacked into Parquet (one row per frame) so the Hugging Face Dataset Viewer and `datasets` work out of the box, with full browsing, search and filtering.
|
| 131 |
+
- RGB, depth (`depth/`), filled depth (`depth_bfx/`), scene category, intrinsics, extrinsics and the final 2D/3D annotations are taken unchanged from the official `SUNRGBD.zip`.
|
| 132 |
+
- Added the 37- and 13-class semantic label maps from Ankur Handa's extraction. Their order follows the official 5,285 / 5,050 train/test split; alignment was verified during the build (label size equals image size for every frame, and unlabeled pixels match each frame's own `seg.mat`).
|
| 133 |
+
- `objects_2d` and `objects_3d` are parsed from `annotation2Dfinal` and `annotation3Dfinal`; the raw JSON is kept in `annotation_2d_json` and `annotation_3d_json`.
|
| 134 |
+
- Not included: the `fullres/` copies, intermediate annotation versions, and room layouts.
|
| 135 |
+
- No images were added, removed, or modified. No labels were changed. The official split is unchanged.
|
| 136 |
+
|
| 137 |
+
---
|
| 138 |
+
|
| 139 |
+
# Dataset Structure
|
| 140 |
+
|
| 141 |
+
| Column | Content |
|
| 142 |
+
|---|---|
|
| 143 |
+
| `image` | RGB image (resolution depends on the sensor) |
|
| 144 |
+
| `depth` | raw depth, 16-bit PNG (official `depth/`) |
|
| 145 |
+
| `depth_filled` | hole-filled depth, 16-bit PNG (official `depth_bfx/`) |
|
| 146 |
+
| `semantic37` | 37-class label map (0 = unlabeled) |
|
| 147 |
+
| `semantic13` | 13-class label map (0 = unknown) |
|
| 148 |
+
| `scene` | scene category (e.g. `bedroom`, `office`) |
|
| 149 |
+
| `sensor`, `source` | `kv1` / `kv2` / `realsense` / `xtion`, and the sub-collection (e.g. `NYUdata`, `b3dodata`, `sun3ddata`) |
|
| 150 |
+
| `path` | original folder inside `SUNRGBD.zip` |
|
| 151 |
+
| `intrinsics` | 3x3 camera matrix, row-major |
|
| 152 |
+
| `extrinsics` | 3x4 rotation/translation from the latest `extrinsics/*.txt` |
|
| 153 |
+
| `objects_2d` | list of `{name, bbox [x, y, w, h], polygon_x, polygon_y}` |
|
| 154 |
+
| `objects_3d` | list of `{name, corners_x, corners_z, y_min, y_max}` (box footprint + height, as in the toolbox) |
|
| 155 |
+
| `annotation_2d_json`, `annotation_3d_json` | raw annotation files |
|
| 156 |
+
|
| 157 |
+
Splits: **train** 5,285 · **test** 5,050 (10,335 total).
|
| 158 |
+
|
| 159 |
+
### Classes
|
| 160 |
+
|
| 161 |
+
**37-class** (`semantic37`, 1..37): wall, floor, cabinet, bed, chair, sofa, table, door, window, bookshelf, picture, counter, blinds, desk, shelves, curtain, dresser, pillow, mirror, floor_mat, clothes, ceiling, books, fridge, tv, paper, towel, shower_curtain, box, whiteboard, person, night_stand, toilet, sink, lamp, bathtub, bag
|
| 162 |
+
|
| 163 |
+
**13-class** (`semantic13`, 1..13): bed, books, ceiling, chair, floor, furniture, objects, picture, sofa, table, tv, wall, window
|
| 164 |
+
|
| 165 |
+
## Usage
|
| 166 |
+
|
| 167 |
+
```python
|
| 168 |
+
from datasets import load_dataset
|
| 169 |
+
|
| 170 |
+
ds = load_dataset("dronefreak/SUN-RGBD", split="train")
|
| 171 |
+
row = ds[0]
|
| 172 |
+
rgb, depth, labels = row["image"], row["depth_filled"], row["semantic37"]
|
| 173 |
+
print(row["scene"], row["sensor"], [o["name"] for o in row["objects_3d"]])
|
| 174 |
+
```
|
| 175 |
+
|
| 176 |
+
---
|
| 177 |
+
|
| 178 |
+
# Dataset Sources
|
| 179 |
+
|
| 180 |
+
## Original Paper
|
| 181 |
+
|
| 182 |
+
**SUN RGB-D: A RGB-D Scene Understanding Benchmark Suite**
|
| 183 |
+
|
| 184 |
+
Shuran Song, Samuel P. Lichtenberg, Jianxiong Xiao
|
| 185 |
+
|
| 186 |
+
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
|
| 187 |
+
|
| 188 |
+
## Citation
|
| 189 |
+
|
| 190 |
+
The authors ask that you cite SUN RGB-D **and** the datasets it builds on:
|
| 191 |
+
|
| 192 |
+
```bibtex
|
| 193 |
+
@inproceedings{song2015sun,
|
| 194 |
+
title = {SUN RGB-D: A RGB-D Scene Understanding Benchmark Suite},
|
| 195 |
+
author = {Song, Shuran and Lichtenberg, Samuel P. and Xiao, Jianxiong},
|
| 196 |
+
booktitle = {CVPR},
|
| 197 |
+
year = {2015}
|
| 198 |
+
}
|
| 199 |
+
@inproceedings{silberman2012indoor,
|
| 200 |
+
title = {Indoor Segmentation and Support Inference from RGBD Images},
|
| 201 |
+
author = {Silberman, Nathan and Hoiem, Derek and Kohli, Pushmeet and Fergus, Rob},
|
| 202 |
+
booktitle = {ECCV},
|
| 203 |
+
year = {2012}
|
| 204 |
+
}
|
| 205 |
+
@inproceedings{janoch2011category,
|
| 206 |
+
title = {A Category-Level 3-D Object Dataset: Putting the Kinect to Work},
|
| 207 |
+
author = {Janoch, Allison and Karayev, Sergey and Jia, Yangqing and Barron, Jonathan T. and Fritz, Mario and Saenko, Kate and Darrell, Trevor},
|
| 208 |
+
booktitle = {ICCV Workshop on Consumer Depth Cameras for Computer Vision},
|
| 209 |
+
year = {2011}
|
| 210 |
+
}
|
| 211 |
+
@inproceedings{xiao2013sun3d,
|
| 212 |
+
title = {SUN3D: A Database of Big Spaces Reconstructed using SfM and Object Labels},
|
| 213 |
+
author = {Xiao, Jianxiong and Owens, Andrew and Torralba, Antonio},
|
| 214 |
+
booktitle = {ICCV},
|
| 215 |
+
year = {2013}
|
| 216 |
+
}
|
| 217 |
+
```
|
| 218 |
+
|
| 219 |
+
---
|
| 220 |
+
|
| 221 |
+
# License
|
| 222 |
+
|
| 223 |
+
The original authors have **not specified a license** for SUN RGB-D; the project page only requires citation of the papers above. This repository therefore uses `license: other` and redistributes the data unofficially for research purposes, with attribution. Users are responsible for ensuring their use complies with any terms the original authors may set. Rights holders can request removal by opening a discussion on this repository.
|
assets/demo_banner.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:abc62fab638cd4243ddcbc1cd0066201a2e3c699dc506c007515327a5b35cb8c
|
| 3 |
+
size 683730
|
assets/demo_banner_poster.jpg
ADDED
|
Git LFS Details
|