Voxel51/SoccerNet-V3

---
annotations_creators: []
language: en
license: mit
size_categories:
- 1K<n<10K
task_categories:
- object-detection
task_ids: []
pretty_name: SoccerNet-V3
tags:
- fiftyone
- group
- object-detection
- sports
- tracking
- action-spotting
- game-state-recognition
dataset_summary: >


This is a [FiftyOne](https://github.com/voxel51/fiftyone) dataset with 1799
samples.

## Installation

If you haven't already, install FiftyOne:

```bash

pip install -U fiftyone

```

## Usage

```python

import fiftyone as fo

import fiftyone.utils.huggingface as fouh

# Load the dataset

# Note: other available arguments include 'max_samples', etc

dataset = fouh.load_from_hub("Voxel51/SoccerNet-V3")

# Launch the App

session = fo.launch_app(dataset)

```
---

# Dataset Card for SoccerNet-V3

SoccerNet is a large-scale dataset for soccer video understanding. It has evolved over the years to include various tasks such as action spotting,
camera calibration, player re-identification and tracking. It is composed of 550 complete broadcast soccer games and 12 single camera games
taken from the major European leagues. SoccerNet is not only dataset, but also yearly challenges where the best teams compete at the international level.


This is a [FiftyOne](https://github.com/voxel51/fiftyone) dataset with 1799 samples.

## Installation

If you haven't already, install FiftyOne:

```bash
pip install -U fiftyone
```

## Usage

```python
import fiftyone as fo
import fiftyone.utils.huggingface as fouh

# Load the dataset
# Note: other available arguments include 'max_samples', etc
dataset = fouh.load_from_hub("Voxel51/SoccerNet-V3")

# Launch the App
session = fo.launch_app(dataset)
```

## Dataset Details

### Dataset Description

<!-- Provide a longer summary of what this dataset is. -->

- **Language(s) (NLP):** en
- **License:** mit

### Dataset Sources
<!-- Provide the basic links for the dataset. -->

- **Repository:** https://github.com/SoccerNet
- **Paper** [SoccerNet 2023 Challenges Results](https://arxiv.org/abs/2309.06006)
- **Demo:** https://try.fiftyone.ai/datasets/soccernet-v3/samples
- **Homepage** https://www.soccer-net.org/

## Dataset Creation

Dataset Authors:

Copyright (c) 2021 holders:

- University of Liège (ULiège), Belgium.
- King Abdullah University of Science and Technology (KAUST), Saudi Arabia.
- Marc Van Droogenbroeck ([e-mail]), Professor at the University of Liège (ULiège).

Code Contributing Authors:

- Anthony Cioppa ([e-mail]), University of Liège (ULiège), Montefiore Institute, TELIM.
- Adrien Deliège ([e-mail]), University of Liège (ULiège), Montefiore Institute, TELIM.
- Silvio Giancola ([e-mail]), King Abdullah University of Science and Technology (KAUST), Image and Video Understanding Laboratory (IVUL), part of the Visual Computing Center (VCC).

Supervision from:

- Bernard Ghanem, King Abdullah University of Science and Technology (KAUST).
- Marc Van Droogenbroeck, University of Liège (ULiège).

### Funding

Anthony Cioppa is funded by the FRIA, Belgium.
This work is supported by the DeepSport and TRAIL projects of the Walloon Region, at the University of Liège (ULiège), Belgium.
This work was supported by the Service Public de Wallonie (SPW) Recherche under the DeepSport project and Grant No.326 2010235 (ARIAC by https://DigitalWallonia4.ai)
This work is also supported by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research (OSR) (award327 OSR-CRG2017-3405).

## Citation

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

```bibtex

@inproceedings{Giancola_2018,
title={SoccerNet: A Scalable Dataset for Action Spotting in Soccer Videos},
url={http://dx.doi.org/10.1109/CVPRW.2018.00223},
DOI={10.1109/cvprw.2018.00223},
booktitle={2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)},
publisher={IEEE},
author={Giancola, Silvio and Amine, Mohieddine and Dghaily, Tarek and Ghanem, Bernard},
year={2018},
month=jun }

@misc{deliège2021soccernetv2,
title={SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos},
author={Adrien Deliège and Anthony Cioppa and Silvio Giancola and Meisam J. Seikavandi and Jacob V. Dueholm and Kamal Nasrollahi and Bernard Ghanem and Thomas B. Moeslund and Marc Van Droogenbroeck},
year={2021},
eprint={2011.13367},
archivePrefix={arXiv},
primaryClass={cs.CV}
}

@misc{cioppa2022soccernettracking,
title={SoccerNet-Tracking: Multiple Object Tracking Dataset and Benchmark in Soccer Videos},
author={Anthony Cioppa and Silvio Giancola and Adrien Deliege and Le Kang and Xin Zhou and Zhiyu Cheng and Bernard Ghanem and Marc Van Droogenbroeck},
year={2022},
eprint={2204.06918},
archivePrefix={arXiv},
primaryClass={cs.CV}
}

@article{Cioppa2022,
title={Scaling up SoccerNet with multi-view spatial localization and re-identification},
author={Cioppa, Anthony and Deli{\`e}ge, Adrien and Giancola, Silvio and Ghanem, Bernard and Van Droogenbroeck, Marc},
journal={Scientific Data},
year={2022},
volume={9},
number={1},
pages={355},
}
```

## Dataset Card Authors

[Jacob Marks](https://huggingface.co/jamarks)

annotations_creators: 无
language: 英语
license: MIT协议
size_categories:
- 1000 < 样本量 < 10000
task_categories:
- 目标检测
task_ids: 无
pretty_name: SoccerNet-V3
tags:
- FiftyOne
- 分组
- 目标检测
- 体育赛事
- 多目标跟踪
- 动作识别
- 比赛状态识别


本数据集为基于FiftyOne（"https://github.com/voxel51/fiftyone"）构建的数据集，共包含1799个样本。

## 安装

若尚未安装FiftyOne，请执行以下命令：
bash
pip install -U fiftyone

## 使用方法
python
import fiftyone as fo
import fiftyone.utils.huggingface as fouh

# 加载数据集
# 注意：其他可用参数包括`max_samples`等
dataset = fouh.load_from_hub("Voxel51/SoccerNet-V3")

# 启动应用
session = fo.launch_app(dataset)

# SoccerNet-V3 数据集卡片

SoccerNet是一款面向足球视频理解的大规模数据集，历经多年迭代，现已涵盖动作识别、相机标定、球员重识别与多目标跟踪等多项任务。该数据集包含来自欧洲顶级联赛的550场完整足球转播赛事与12场单摄像头赛事。SoccerNet不仅是数据集，更是每年举办的国际性挑战赛，汇聚全球顶尖团队同台竞技。


本数据集为基于FiftyOne（"https://github.com/voxel51/fiftyone"）构建的数据集，共包含1799个样本。

## 数据集详情

### 数据集概述
<!-- 请提供该数据集的详细摘要 -->
- **语言（自然语言处理）：** 英语
- **许可证：** MIT协议

### 数据集来源
<!-- 请提供数据集的基础链接 -->
- **代码仓库：** https://github.com/SoccerNet
- **论文：** [SoccerNet 2023 挑战赛结果]("https://arxiv.org/abs/2309.06006")
- **演示站点：** https://try.fiftyone.ai/datasets/soccernet-v3/samples
- **官方主页：** https://www.soccer-net.org/

## 数据集构建
数据集作者：
版权所有 (c) 2021 持有者：
- 比利时列日大学（University of Liège, ULiège）
- 沙特阿拉伯阿卜杜拉国王科技大学（King Abdullah University of Science and Technology, KAUST）
- Marc Van Droogenbroeck（[e-mail]），列日大学（ULiège）教授。

代码贡献作者：
- Anthony Cioppa（[e-mail]），列日大学（ULiège）蒙特菲奥雷研究所TELIM实验室
- Adrien Deliège（[e-mail]），列日大学（ULiège）蒙特菲奥雷研究所TELIM实验室
- Silvio Giancola（[e-mail]），阿卜杜拉国王科技大学（KAUST）图像与视频理解实验室（Image and Video Understanding Laboratory, IVUL），隶属于视觉计算中心（Visual Computing Center, VCC）。

指导教师：
- Bernard Ghanem，阿卜杜拉国王科技大学（KAUST）
- Marc Van Droogenbroeck，列日大学（ULiège）

### 资助信息
Anthony Cioppa 受比利时FRIA项目资助。
本研究受比利时列日大学（ULiège）瓦隆大区（Walloon Region）下属DeepSport与TRAIL项目资助。
本研究受瓦隆公共服务部门（Service Public de Wallonie, SPW）研究部资助，项目编号为326 2010235（由https://DigitalWallonia4.ai提供的ARIAC项目）。
本研究同时获得阿卜杜拉国王科技大学（KAUST）赞助研究办公室（Office of Sponsored Research, OSR）资助（项目编号OSR-CRG2017-3405，编号327）。

## 引用
<!-- 若该数据集对应相关论文或博客，请在此处提供APA与BibTeX格式的引用信息 -->
**BibTeX格式：**
bibtex
@inproceedings{Giancola_2018,
title={SoccerNet: A Scalable Dataset for Action Spotting in Soccer Videos},
url={http://dx.doi.org/10.1109/CVPRW.2018.00223},
DOI={10.1109/cvprw.2018.00223},
booktitle={2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)},
publisher={IEEE},
author={Giancola, Silvio and Amine, Mohieddine and Dghaily, Tarek and Ghanem, Bernard},
year={2018},
month=jun
}

@misc{deliège2021soccernetv2,
title={SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos},
author={Adrien Deliège and Anthony Cioppa and Silvio Giancola and Meisam J. Seikavandi and Jacob V. Dueholm and Kamal Nasrollahi and Bernard Ghanem and Thomas B. Moeslund and Marc Van Droogenbroeck},
year={2021},
eprint={2011.13367},
archivePrefix={arXiv},
primaryClass={cs.CV}
}

@misc{cioppa2022soccernettracking,
title={SoccerNet-Tracking: Multiple Object Tracking Dataset and Benchmark in Soccer Videos},
author={Anthony Cioppa and Silvio Giancola and Adrien Deliege and Le Kang and Xin Zhou and Zhiyu Cheng and Bernard Ghanem and Marc Van Droogenbroeck},
year={2022},
eprint={2204.06918},
archivePrefix={arXiv},
primaryClass={cs.CV}
}

@article{Cioppa2022,
title={Scaling up SoccerNet with multi-view spatial localization and re-identification},
author={Cioppa, Anthony and Deliège, Adrien and Giancola, Silvio and Ghanem, Bernard and Van Droogenbroeck, Marc},
journal={Scientific Data},
year={2022},
volume={9},
number={1},
pages={355},
}

## 数据集卡片作者
[Jacob Marks]("https://huggingface.co/jamarks")

原始信息汇总

数据集概述

• 名称: SoccerNet-V3
• 语言: 英语 (en)
• 许可: MIT
• 大小: 1K<n<10K
• 任务类别: 目标检测 (object-detection)
• 标签: fiftyone, group, object-detection, sports, tracking, action-spotting, game-state-recognition

数据集描述

SoccerNet-V3 是一个用于足球视频理解的大型数据集，包含550场完整的广播足球比赛和12场单摄像机比赛，来自主要的欧洲联赛。该数据集不仅用于数据分析，还用于年度国际挑战赛。

数据集来源

• 仓库: https://github.com/SoccerNet
• 论文: SoccerNet 2023 Challenges Results
• 演示: https://try.fiftyone.ai/datasets/soccernet-v3/samples
• 主页: https://www.soccer-net.org/

数据集创建

• 作者:
  • 大学: 列日大学 (ULiège), 沙特阿拉伯阿卜杜拉国王科技大学 (KAUST)
  • 个人: Marc Van Droogenbroeck, Anthony Ciopma, Adrien Deliège, Silvio Giancola
• 监督: Bernard Ghanem, Marc Van Droogenbroeck

资金支持

• 由FRIA, Belgium, 瓦隆地区DeepSport和TRAIL项目, 以及KAUST Office of Sponsored Research (OSR) 支持。

引用信息

bibtex @inproceedings{Giancola_2018, title={SoccerNet: A Scalable Dataset for Action Spotting in Soccer Videos}, url={http://dx.doi.org/10.1109/CVPRW.2018.00223}, DOI={10.1109/cvprw.2018.00223}, booktitle={2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)}, publisher={IEEE}, author={Giancola, Silvio and Amine, Mohieddine and Dghaily, Tarek and Ghanem, Bernard}, year={2018}, month=jun }

@misc{deliège2021soccernetv2, title={SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos}, author={Adrien Deliège and Anthony Cioppa and Silvio Giancola and Meisam J. Seikavandi and Jacob V. Dueholm and Kamal Nasrollahi and Bernard Ghanem and Thomas B. Moeslund and Marc Van Droogenbroeck}, year={2021}, eprint={2011.13367}, archivePrefix={arXiv}, primaryClass={cs.CV} }

@misc{cioppa2022soccernettracking, title={SoccerNet-Tracking: Multiple Object Tracking Dataset and Benchmark in Soccer Videos}, author={Anthony Cioppa and Silvio Giancola and Adrien Deliege and Le Kang and Xin Zhou and Zhiyu Cheng and Bernard Ghanem and Marc Van Droogenbroeck}, year={2022}, eprint={2204.06918}, archivePrefix={arXiv}, primaryClass={cs.CV} }

@article{Cioppa2022, title={Scaling up SoccerNet with multi-view spatial localization and re-identification}, author={Cioppa, Anthony and Deli{`e}ge, Adrien and Giancola, Silvio and Ghanem, Bernard and Van Droogenbroeck, Marc}, journal={Scientific Data}, year={2022}, volume={9}, number={1}, pages={355}, }

搜集汇总

数据集介绍

构建方式

SoccerNet-V3数据集是由比利时列日大学和沙特阿拉伯国王阿卜杜拉科技大学的研究人员共同构建的，它包含了550场完整的欧洲主要联赛的足球比赛视频和12场单摄像头游戏视频。该数据集的构建采用了先进的视频理解技术，对足球比赛中的动作进行标注和分类，旨在推动足球视频理解领域的研究。

特点

SoccerNet-V3数据集的特点在于其规模宏大、多样性丰富，并包含了多种任务类型，如动作检测、摄像头校准、球员重识别和跟踪等。它不仅是一个数据集，还伴随着年度挑战，吸引了国际上的顶尖团队参与竞争。此外，该数据集遵循MIT许可证，保证了其使用的开放性和灵活性。

使用方法

使用SoccerNet-V3数据集首先需要安装FiftyOne库，通过pip命令即可完成。加载数据集时，可以利用fiftyone.utils.huggingface模块中的load_from_hub函数。数据集加载后，可以通过FiftyOne的应用程序界面进行交互式探索，方便研究人员进行数据分析和模型训练。

背景与挑战

背景概述

SoccerNet-V3数据集，由比利时列日大学（University of Liège）和沙特阿拉伯国王阿卜杜拉科技大学（King Abdullah University of Science and Technology）的研究人员共同创建，旨在推动足球视频理解领域的研究。该数据集自推出以来，不断演化，涵盖了动作检测、摄像头校准、球员重识别与跟踪等多种任务。它由550场完整的广播足球比赛和12场单一摄像头游戏组成，主要来自欧洲主要联赛。SoccerNet不仅是一个数据集，还每年举办国际级别的挑战赛，吸引顶尖团队竞技。SoccerNet-V3数据集包含1799个样本，采用MIT许可证发布，支持FiftyOne数据集管理工具。

当前挑战

在研究领域，SoccerNet-V3数据集面临的挑战包括如何更准确地定位和识别足球比赛中的动作和事件，以及如何处理多视角的视频数据。构建过程中，研究者们遇到的挑战包括如何高效地从大量视频数据中提取有用信息，确保标注的质量和一致性，以及如何设计能够适应不同比赛场景和摄像角度的算法。此外，数据集的规模和复杂性也对计算资源和算法效率提出了更高的要求。

常用场景

经典使用场景

在体育视频分析领域，SoccerNet-V3数据集凭借其丰富的标注和多样化的任务类型，成为了一项不可或缺的资源。该数据集的经典使用场景在于对足球比赛视频进行动作定位、球员跟踪和游戏状态识别，进而为用户提供详尽的比赛分析。

衍生相关工作

基于SoccerNet-V3数据集，已经衍生出多项相关工作，包括对数据集的扩展、新的标注策略、以及更先进的动作识别和跟踪算法的研究。这些工作进一步推动了体育视频分析领域的发展，为相关技术的商业化应用奠定了基础。

数据集最近研究

最新研究方向

在体育视频理解领域，SoccerNet-V3数据集正引领着前沿研究方向。该数据集不仅包含了大规模的足球比赛视频，还涵盖了动作定位、相机校准、球员重识别与追踪等多种任务。近期研究聚焦于利用深度学习技术提升对足球比赛视频中复杂动作的检测与识别，以及对比赛状态的精准判别。这些研究对于提高体育视频分析的性能，以及为观众提供更为丰富的观看体验具有重要意义。通过SoccerNet-V3，学者们能够进一步探索视频内容理解的多维度挑战，为智能体育分析领域的发展贡献力量。