Datasets:
annotations_creators:
- crowdsourced
language:
- en
language_creators:
- expert-generated
license:
- cc-by-nc-4.0
multilinguality:
- monolingual
pretty_name: SportsHHI
size_categories: []
source_datasets:
- original
tags:
- video
- video relation detection
- human-human interaction detection
task_categories:
- image-classification
- object-detection
- other
task_ids:
- multi-class-image-classification
extra_gated_heading: Acknowledge license to accept the repository
extra_gated_prompt: >-
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0
International License
extra_gated_fields:
Institute: text
I want to use this dataset for:
type: select
options:
- Research
- Education
- label: Other
value: other
I agree to use this dataset for non-commerical use ONLY: checkbox
Dataset Card for SportsHHI
Table of Contents
Dataset Description
- Repository: https://github.com/MCG-NJU/SportsHHI
- Paper: https://arxiv.org/abs/2404.04565
- Point of Contact: mailto: [email protected]
Dataset Summary
SportsHHI is a dataset for video human-human interaction detection. Please refer to SportsHHI: A Dataset for Human-Human Interaction Detection in Sports Videos for more details. Please refer to this repository for training and evaluation.
Supported Tasks and Leaderboards
Human-Human Interaction Detection
Details about training and evaluation can be found in the GitHub Repository.
Languages
The class labels in the dataset are in English.
Dataset Structure
The dataset contains frames.zip
and annotations
.
frames.zip
contains extracted video frames.
annotations
contains annotaions about human boxes, interactions, and interaction classes.
If you have any network issues, please refer to pan.txt
for data downloading.
Data Splits
train | validation | |
---|---|---|
# of tubes | 38527 | 12122 |
Dataset Creation
Curation Rationale
comprehending high-level inter- actions between humans is crucial for understanding com- plex multi-person videos, such as sports and surveillance videos. To address this issue, authors propose a new video visual relation detection task: video human-human interaction detection, and build a dataset named SportsHHI for it.
Source Data
Initial Data Collection and Normalization
We carefully selected 80 basketball and 80 volleyball videos from the MultiSports dataset to cover various types of games including men’s, women’s, national team, and club games. The average length of the videos is 603 frames and the frame rate of the videos is 25FPS. All videos have a high resolution of 720P.
Annotations
Please refer to SportsHHI: A Dataset for Human-Human Interaction Detection in Sports Videos for more information.
Additional Information
Dataset Curators
Authors of SportsHHI: A Dataset for Human-Human Interaction Detection in Sports Videos
Tao Wu
Runyu He
Gangshan Wu
Limin Wang
Licensing Information
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Citation Information
If you find this dataset useful, please cite as
@misc{wu2024sportshhi,
title={SportsHHI: A Dataset for Human-Human Interaction Detection in Sports Videos},
author={Tao Wu and Runyu He and Gangshan Wu and Limin Wang},
year={2024},
eprint={2404.04565},
archivePrefix={arXiv},
primaryClass={cs.CV}
}