Korean_Warning_Labels_Dataset

# Korean Warning Labels Dataset
*This dataset contains high-resolution images of Korean warning and safety labels, including product hazard signs, electrical warnings, chemical safety labels, and public safety notices. It supports AI research in OCR, text detection, and safety compliance analysis.*

## Contact
For queries or collaborations related to this dataset, contact:
- [e-mail]
- [e-mail]

## Supported Tasks

- **Task Categories**:
- Image Classification
- Text Recognition (OCR)
- Scene Understanding

- **Supported Tasks**:
- Korean text extraction from warning labels and safety signs
- Classification of warning types (hazard, electrical, chemical, public)
- Detection of iconography and visual symbols associated with warnings
- Multilingual OCR for labels including Korean-English text
- AI modeling for automated safety compliance and hazard recognition

## Languages

- **Primary Language**: Korean
- **Secondary Presence**: English (on bilingual labels)

## Dataset Creation

### Curation Rationale
This dataset was curated to facilitate AI systems that can automatically interpret safety and warning information in Korean contexts. It enables research in OCR, computer vision, and automated safety compliance applications.

### Source Data
- **Contributors**: Field collection from products, public signage, industrial environments, and safety documentation
- **Collection Process**: Images were captured of public warning signs, product labels, and instructional materials. Personal identifiers were removed and sensitive information excluded.

### Other Known Limitations
- **Bias**: Urban and industrial labels are more represented than rural or informal warning signs
- **Visual Variability**: Differences in lighting, angles, and label wear may affect OCR accuracy
- **Content Scope**: Focused on textual and symbolic warnings; contextual safety information may be limited

## Intended Uses

### ✅ Direct Use
- Training OCR and visual recognition models for safety labels
- Automated hazard detection and compliance monitoring
- Research in human-machine interaction and safety signage understanding
- Multimodal AI applications combining text and symbols for safety

### ❌ Out-of-Scope Use
- Tracking or surveillance of individuals via safety signage
- Commercial reuse of proprietary safety designs without proper permissions
- Misuse of dataset for non-safety-related commercial purposes

## License

CC BY 4.0

# 韩语警示标签数据集
*本数据集收录了韩语警示与安全标签的高分辨率图像，涵盖产品危险标识、电气警示、化工安全标签及公共安全告示，可支撑光学字符识别（Optical Character Recognition）、文本检测及安全合规分析相关的AI研究。

## 联系方式
如需咨询该数据集相关事宜或寻求合作，请联系：
- [e-mail]
- [e-mail]

## 支持任务

- **任务类别**：
- 图像分类
- 文本识别（Optical Character Recognition）
- 场景理解

- **支持任务**：
- 从警示标签与安全标识中提取韩语文本
- 警示类型分类（危险警示、电气警示、化工安全警示、公共安全警示）
- 关联警示的图像与视觉符号检测
- 支持含韩英双语文本的标签多语言OCR
- 用于自动化安全合规与危险识别的AI建模

## 语言分布

- **主要语言**：韩语
- **次要语言**：双语标签中包含英语

## 数据集构建

### 遴选依据
本数据集的遴选旨在助力能够自动解读韩语场景下安全与警示信息的AI系统，为光学字符识别、计算机视觉及自动化安全合规应用的研究提供支撑。

### 源数据
- **数据贡献者**：从产品、公共标识、工业环境及安全文档中实地采集
- **采集流程**：采集公共警示标识、产品标签与安全指导材料的图像，已移除个人识别信息并排除敏感内容。

### 已知其他局限性
- **偏差**：城市与工业场景的标签占比高于农村或非正规警示标识
- **视觉差异性**：光照、拍摄角度及标签磨损程度的差异可能影响OCR识别精度
- **内容范围**：仅聚焦文本与符号类警示，上下文安全信息的覆盖可能有限

## 预期用途

### ✅ 直接用途
- 训练面向安全标签的OCR与视觉识别模型
- 自动化危险识别与合规监测
- 人机交互与安全标识理解相关研究
- 面向安全场景的文本与符号融合多模态AI应用

### ❌ 超出范围用途
- 通过安全标识对个体进行追踪或监视
- 未经许可商用复用受版权保护的安全设计
- 将数据集用于非安全相关的商业用途

## 许可证

CC BY 4.0