Rename VCBench -> SVCBench; fix license to CC BY-NC 4.0; real BibTeX & links
Browse files
README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
---
|
| 2 |
-
license:
|
| 3 |
task_categories:
|
| 4 |
- video-classification
|
| 5 |
- question-answering
|
|
@@ -10,15 +10,22 @@ tags:
|
|
| 10 |
- temporal-reasoning
|
| 11 |
- counting
|
| 12 |
- benchmark
|
|
|
|
| 13 |
size_categories:
|
| 14 |
- 1K<n<10K
|
|
|
|
| 15 |
---
|
| 16 |
|
| 17 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
## Dataset Description
|
| 20 |
|
| 21 |
-
This
|
| 22 |
|
| 23 |
### Dataset Summary
|
| 24 |
|
|
@@ -66,7 +73,7 @@ Videos are clipped from multiple source datasets:
|
|
| 66 |
- YouTube walking tours and sports videos
|
| 67 |
- RoomTour3D (indoor navigation)
|
| 68 |
- Ego4D (first-person view)
|
| 69 |
-
- ScanNet,
|
| 70 |
- TOMATO, CODa, OmniWorld (temporal reasoning)
|
| 71 |
- Simulated physics videos
|
| 72 |
|
|
@@ -80,7 +87,7 @@ import cv2
|
|
| 80 |
|
| 81 |
# Download a specific video
|
| 82 |
video_path = hf_hub_download(
|
| 83 |
-
repo_id="
|
| 84 |
filename="e1action_0000_00.mp4",
|
| 85 |
repo_type="dataset"
|
| 86 |
)
|
|
@@ -96,12 +103,12 @@ cap = cv2.VideoCapture(video_path)
|
|
| 96 |
pip install huggingface_hub
|
| 97 |
|
| 98 |
# Download entire dataset
|
| 99 |
-
huggingface-cli download
|
| 100 |
```
|
| 101 |
|
| 102 |
## Annotations
|
| 103 |
|
| 104 |
-
For complete annotations including questions, query points, and ground truth answers, please refer to the
|
| 105 |
- Object counting annotations: `object_count_data/*.json`
|
| 106 |
- Event counting annotations: `event_counting_data/*.json`
|
| 107 |
|
|
@@ -123,20 +130,25 @@ All videos have been validated for:
|
|
| 123 |
|
| 124 |
## Citation
|
| 125 |
|
| 126 |
-
If you use this dataset, please cite the
|
| 127 |
|
| 128 |
```bibtex
|
| 129 |
-
@
|
| 130 |
-
title={
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
|
|
|
|
|
|
|
|
|
| 134 |
}
|
| 135 |
```
|
| 136 |
|
| 137 |
## License
|
| 138 |
|
| 139 |
-
|
|
|
|
|
|
|
| 140 |
|
| 141 |
## Dataset Statistics
|
| 142 |
|
|
@@ -154,4 +166,4 @@ MIT License - See LICENSE file for details.
|
|
| 154 |
|
| 155 |
## Contact
|
| 156 |
|
| 157 |
-
For questions or issues, please open an issue in the
|
|
|
|
| 1 |
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
task_categories:
|
| 4 |
- video-classification
|
| 5 |
- question-answering
|
|
|
|
| 10 |
- temporal-reasoning
|
| 11 |
- counting
|
| 12 |
- benchmark
|
| 13 |
+
- streaming
|
| 14 |
size_categories:
|
| 15 |
- 1K<n<10K
|
| 16 |
+
pretty_name: SVCBench
|
| 17 |
---
|
| 18 |
|
| 19 |
+
# SVCBench: Streaming Video Counting Benchmark
|
| 20 |
+
|
| 21 |
+
This dataset contains the **clipped video segments** for **SVCBench**, a Streaming Video Counting Benchmark for Spatial-Temporal State Maintenance. It repositions counting as a minimal, controlled probe for diagnosing how video understanding models maintain world state along the video timeline.
|
| 22 |
+
|
| 23 |
+
- **Project Page**: https://buaa-colalab.github.io/SVCBench/
|
| 24 |
+
- **Code**: https://github.com/buaa-colalab/SVCBench
|
| 25 |
|
| 26 |
## Dataset Description
|
| 27 |
|
| 28 |
+
This repository provides **4,574 clipped video segments** used by SVCBench, designed for evaluating spatial-temporal state maintenance capabilities in video understanding models.
|
| 29 |
|
| 30 |
### Dataset Summary
|
| 31 |
|
|
|
|
| 73 |
- YouTube walking tours and sports videos
|
| 74 |
- RoomTour3D (indoor navigation)
|
| 75 |
- Ego4D (first-person view)
|
| 76 |
+
- ScanNet, ScanNet++, ARKitScenes (3D indoor scenes)
|
| 77 |
- TOMATO, CODa, OmniWorld (temporal reasoning)
|
| 78 |
- Simulated physics videos
|
| 79 |
|
|
|
|
| 87 |
|
| 88 |
# Download a specific video
|
| 89 |
video_path = hf_hub_download(
|
| 90 |
+
repo_id="buaaplay/SVCBench",
|
| 91 |
filename="e1action_0000_00.mp4",
|
| 92 |
repo_type="dataset"
|
| 93 |
)
|
|
|
|
| 103 |
pip install huggingface_hub
|
| 104 |
|
| 105 |
# Download entire dataset
|
| 106 |
+
huggingface-cli download buaaplay/SVCBench --repo-type dataset --local-dir ./svcbench_videos
|
| 107 |
```
|
| 108 |
|
| 109 |
## Annotations
|
| 110 |
|
| 111 |
+
For complete annotations including questions, query points, and ground truth answers, please refer to the SVCBench code repository (https://github.com/buaa-colalab/SVCBench):
|
| 112 |
- Object counting annotations: `object_count_data/*.json`
|
| 113 |
- Event counting annotations: `event_counting_data/*.json`
|
| 114 |
|
|
|
|
| 130 |
|
| 131 |
## Citation
|
| 132 |
|
| 133 |
+
If you use this dataset, please cite the SVCBench paper:
|
| 134 |
|
| 135 |
```bibtex
|
| 136 |
+
@inproceedings{liu2026svcbench,
|
| 137 |
+
title = {SVCBench: A Streaming Video Counting Benchmark for
|
| 138 |
+
Spatial-Temporal State Maintenance},
|
| 139 |
+
author = {Liu, Pengyiang and Shi, Zhongyue and Hao, Hongye and Fu, Qi and
|
| 140 |
+
Bi, Xueting and Zhang, Siwei and Hu, Xiaoyang and Wang, Zitian and
|
| 141 |
+
Huang, Linjiang and Liu, Si},
|
| 142 |
+
booktitle = {European Conference on Computer Vision (ECCV)},
|
| 143 |
+
year = {2026}
|
| 144 |
}
|
| 145 |
```
|
| 146 |
|
| 147 |
## License
|
| 148 |
|
| 149 |
+
The SVCBench **annotations** and our **self-generated physics-simulation videos** are released under the **Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)** license, for **non-commercial research use only**.
|
| 150 |
+
|
| 151 |
+
The clipped video segments are derived from **third-party sources** (YouTube, Ego4D, ScanNet, ScanNet++, ARKitScenes, RoomTour3D, TOMATO, CODa, OmniWorld). These clips remain subject to the **original licenses and terms of use of their respective source datasets**, and are provided here solely for non-commercial academic research. Users are responsible for reviewing and complying with each source dataset's license and terms. If you are a rights holder with any concern about specific content, please open an issue and we will address it promptly.
|
| 152 |
|
| 153 |
## Dataset Statistics
|
| 154 |
|
|
|
|
| 166 |
|
| 167 |
## Contact
|
| 168 |
|
| 169 |
+
For questions or issues, please open an issue in the [SVCBench repository](https://github.com/buaa-colalab/SVCBench).
|