Underwater Drowning Detection Dataset

<b>Underwater Drowning Detection Dataset</b><br>This dataset contains 5,613 manually annotated underwater images for drowning detection research, captured in controlled swimming pool environments. It provides a balanced distribution of three behavioral states:<br><b>Swimming</b> (1,871 images)<b>Struggling</b> (1,871 images)<b>Drowning</b> (1,871 images)All images were collected under real underwater conditions and annotated for object detection tasks using the YOLO format.<b>Key Features</b>High-resolution underwater images (640×640 pixels, RGB)YOLO <code>.txt</code> annotations with bounding boxes for three behavior classesBalanced class distribution to minimize model biasData collected ethically with lifeguard supervision and participant consentIncludes realistic challenges such as water distortion and lighting variability<b>Technical Details</b><b>Total Images:</b> 5,613<b>Training/Validation Split:</b> 4,488 / 1,125<b>Classes:</b> Swimming, Struggling, Drowning<b>Format:</b> JPEG + YOLO annotation files<b>Resolution:</b> 640×640 pixels<b>Baseline Performance:</b> YOLOv8n achieved 97.5% mAP@50 on this dataset<b>Annotation Format</b><br>Each image has a corresponding .txt file with annotations in YOLO format, where each line follows this structure: <br><br><b>Field Descriptions:</b><br><br>class_id: Integer label for the class<br>0 = Swimming<br>1 = Struggling<br>2 = Drowningx_center, y_center: Normalized center coordinates of the bounding box (values between 0.0 and 1.0)width, height: Normalized width and height of the bounding box (values between 0.0 and 1.0)<b>Example Annotation:</b><br><br>0 0.509896 0.568519 0.453125 0.581481This line indicates a “Swimming” detection (class_id = 0) with a bounding box centered at 50.99% (horizontal) and 56.85% (vertical) of the image dimensions, covering 45.31% of the width and 58.15% of the height.<b>Dataset Folder Structure</b><br>datasets/<br>├── images/<br>│ ├── train/<br>│ │ ├── frame_00001.jpg<br>│ │ └── ...<br>│ └── val/<br>│ ├── frame_04489.jpg<br>│ └── ...<br>│<br>├── labels/<br>│ ├── train/<br>│ │ ├── frame_00001.txt<br>│ │ └── ...<br>│ └── val/<br>│ ├── frame_04489.txt<br>│ └── ...<br>│<br>├── classes.txt<br>├── README.md<br><b>Use and Applications</b><br>This dataset is designed to support the development and evaluation of real-time AI systems for aquatic safety, including:<br>Drowning detection modelsMulti-class object detection in underwater environmentsResearch in underwater computer vision and human activity recognition<b>Citation</b><br>If you use this dataset, please cite:<br><pre>graphqlCopyEdit<pre>@dataset{underwater_drowning_detection_2025,<br> title = {Underwater Drowning Detection Dataset},<br> author = {Hamad Alzaabi and Saif Alzaabi and Sarah Kohail},<br> year = {2025},<br> publisher = {Figshare},<br> note = {Manually annotated underwater images for drowning detection research}<br>}<br></pre></pre>Please also cite the related publication:<br><br>mathematicaCopyEdit<pre>@inproceedings{Alzaabi2025,<br> author = {Hamad Alzaabi and Saif Alzaabi and Sarah Kohail},<br> title = {Multi‑Swimmer Drowning Detection Using a Custom Annotated Underwater Dataset and Real‑Time AI},<br> booktitle = {Proceedings of the International Conference on Image Analysis and Processing (ICIAP)},<br> year = {2025}<br>}</pre><br>

**水下溺水检测数据集**
本数据集包含5613张经人工标注的水下图像，用于溺水检测研究，所有图像均采集于受控游泳池环境中。数据集涵盖三类行为状态且样本分布均衡：
- **游泳**：1871张
- **挣扎**：1871张
- **溺水**：1871张
所有图像均采集自真实水下场景，并采用YOLO（You Only Look Once）格式进行目标检测任务标注。

### 核心特性
1. 高分辨率水下RGB图像，分辨率为640×640像素；
2. 针对三类行为的边界框标注采用YOLO格式的`.txt`文件存储；
3. 类别分布均衡，可有效降低模型偏差；
4. 数据集采集流程符合伦理规范，全程有救生员监督并获得参与者知情同意；
5. 涵盖真实水下场景挑战，如水体畸变与光照变化。

### 技术细节
- 总图像数：5613
- 训练集/验证集划分：4488张 / 1125张
- 类别：游泳、挣扎、溺水
- 存储格式：JPEG图像 + YOLO标注文件
- 图像分辨率：640×640像素
- 基准性能：YOLOv8n在本数据集上实现了97.5%的mAP@50（交并比阈值为0.5时的平均精度均值）指标。

### 标注格式
每张图像对应一个`.txt`格式的标注文件，标注采用YOLO格式，每行遵循以下结构：

#### 字段说明
- `class_id`：类别整数标签
- 0 = 游泳
- 1 = 挣扎
- 2 = 溺水
- `x_center`、`y_center`：边界框中心的归一化坐标（取值范围为0.0至1.0）
- `width`、`height`：边界框的归一化宽高（取值范围为0.0至1.0）

#### 标注示例
`0 0.509896 0.568519 0.453125 0.581481`
该行表示一个“游泳”类目标（`class_id=0`），边界框中心位于图像横向50.99%、纵向56.85%的位置，框宽占图像宽度的45.31%，框高占图像高度的58.15%。

### 数据集文件夹结构

datasets/
├── images/
│ ├── train/
│ │ ├── frame_00001.jpg
│ │ └── ...
│ └── val/
│ ├── frame_04489.jpg
│ └── ...
│
├── labels/
│ ├── train/
│ │ ├── frame_00001.txt
│ │ └── ...
│ └── val/
│ ├── frame_04489.txt
│ └── ...
│
├── classes.txt
├── README.md

### 应用场景
本数据集旨在支持水上安全实时AI系统的开发与评估，具体应用包括：
1. 溺水检测模型研发；
2. 水下环境多类别目标检测研究；
3. 水下计算机视觉与人类行为识别相关研究。

### 引用规范
若您使用本数据集，请引用以下文献：
graphqlCopyEdit
@dataset{underwater_drowning_detection_2025,
title = {Underwater Drowning Detection Dataset},
author = {Hamad Alzaabi and Saif Alzaabi and Sarah Kohail},
year = {2025},
publisher = {Figshare},
note = {Manually annotated underwater images for drowning detection research}
}

同时请引用相关会议论文：
mathematicaCopyEdit
@inproceedings{Alzaabi2025,
author = {Hamad Alzaabi and Saif Alzaabi and Sarah Kohail},
title = {Multi‑Swimmer Drowning Detection Using a Custom Annotated Underwater Dataset and Real‑Time AI},
booktitle = {Proceedings of the International Conference on Image Analysis and Processing (ICIAP)},
year = {2025}
}

搜集汇总

背景与挑战

背景概述

该数据集是一个专门用于溺水检测研究的水下图像数据集，包含5,613张手动标注的图像，均匀覆盖游泳、挣扎和溺水三种行为状态，图像分辨率为640×640像素，并采用YOLO格式进行标注。数据集在受控游泳池环境中采集，具有水失真和光照变化等现实挑战，已划分为训练集和验证集，支持开发实时AI系统以提升水上安全。