MEDIC

# MEDIC: A Multi-Task Learning Dataset for Disaster Image Classification

## Data

The MEDIC is the largest multi-task learning disaster-related dataset, an extended version of the crisis image benchmark dataset. It consists of data from several sources, including CrisisMMD, data from AIDR, and the Damage Multimodal Dataset (DMD). The dataset contains 71,198 images.

## Table of Contents

- [Data format and directories](#data-format-and-directories)
- [Disaster response tasks](#disaster-response-tasks)
- [Downloads](#downloads)
- [Citation](#citation)
- [Terms of Use](#terms-of-use)

## Data Format and Directories

### Directories

- **data**: Main directory with the following subdirectories:
- **aidr_disaster_types/**: Contains images collected using AIDR system for disaster types task.
- **aidr_info/**: Contains images collected using AIDR system for informativeness task.
- **ASONAM17_Damage_Image_Dataset/**: Damage Assessment Dataset.
- **crisismmd/**: CrisisMMD dataset.
- **multimodal-deep-learning-disaster-response-mouzannar/**: Damage Multimodal Dataset (DMD).
- **MEDIC_train.tsv, MEDIC_dev.tsv, MEDIC_test.tsv**: Training, development, and testing files with specific file formats.
- **LICENSE_CC_BY_NC_SA_4.0.txt**: License information.
- **terms-of-use.txt**: Terms and conditions.

### Format

- **image_id**: Corresponds to the tweet ID from Twitter or ID from the respective source.
- **event_name**: Name of the event or data source.
- **image_path**: Relative path of the image.
- **damage_severity**: Damage severity class label.
- **informative**: Informativeness class label.
- **humanitarian**: Humanitarian class label.
- **disaster_types**: Disaster types class label.

## Disaster Response Tasks

1. **Disaster Types**
- Earthquake
- Fire
- Flood
- Hurricane
- Landslide
- Not disaster
- Other disaster

2. **Informativeness**
- Informative
- Not informative

3. **Humanitarian Categories**
- Affected, injured, or dead people
- Infrastructure and utility damage
- Not humanitarian
- Rescue, volunteering, or donation effort

4. **Damage Severity Assessment**
- Little or no damage
- Mild damage
- Severe damage

## Downloads

- **MEDIC Dataset, version v1.0**: [Download](https://crisisnlp.qcri.org/data/medic/MEDIC.tar.gz) (11 GB)
- **Code**: [GitHub Repository](https://github.com/firojalam/medic)

### License
The MEDIC dataset is published under CC BY-NC-SA 4.0 license, which means everyone can use this dataset for non-commercial research purpose: https://creativecommons.org/licenses/by-nc/4.0/.
See LICENSE_CC_BY_NC_SA_4.0.txt

## Citation

Please cite the following papers if you use this dataset in your research:

1. Firoj Alam, Tanvirul Alam, Md. Arid Hasan, Abul Hasnat, Muhammad Imran, Ferda Ofli. *MEDIC: A Multi-Task Learning Dataset for Disaster Image Classification.* Neural Computing and Applications, 35(3):2609–2632, 2023. [paper](https://link.springer.com/content/pdf/10.1007/s00521-022-07717-0.pdf) [Arxiv](https://arxiv.org/pdf/2108.12828)
2. Firoj Alam, Ferda Ofli, Muhammad Imran, Tanvirul Alam, Umair Qazi. *Deep Learning Benchmarks and Datasets for Social Media Image Classification for Disaster Response.* In 2020 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), 2020.
3. Firoj Alam, Ferda Ofli, and Muhammad Imran. *CrisisMMD: Multimodal Twitter Datasets from Natural Disasters.* In Proceedings of the 12th International AAAI Conference on Web and Social Media (ICWSM), 2018, Stanford, California, USA.
4. Hussein Mozannar, Yara Rizk, and Mariette Awad. *Damage Identification in Social Media Posts using Multimodal Deep Learning.* In Proc. of ISCRAM, May 2018, pp. 529–543.
5. Dat Tien Nguyen, Ferda Ofli, Muhammad Imran, and Prasenjit Mitra. *Damage assessment from social media imagery data during disasters.* In Proc. of ASONAM, pages 1–8, Aug 2017.

```
@article{alam2022medic,
title={{MEDIC}: A Multi-Task Learning Dataset for Disaster Image Classification},
author={Firoj Alam and Tanvirul Alam and Md. Arid Hasan and Abul Hasnat and Muhammad Imran and Ferda Ofli},
Keywords = {Multi-task Learning, Social media images, Image Classification, Natural disasters, Crisis Informatics, Deep learning, Dataset},
journal={Neural Computing and Applications},
volume={35},
issue={3},
pages={2609--2632},
year={2023},
publisher={Springer}
}

@InProceedings{crisismmd2018icwsm,
author = {Alam, Firoj and Ofli, Ferda and Imran, Muhammad},
title = {{CrisisMMD}: Multimodal Twitter Datasets from Natural Disasters},
booktitle = {Proceedings of the 12th International AAAI Conference on Web and Social Media (ICWSM)},
year = {2018},
month = {June},
date = {23-28},
location = {USA}
}

@inproceedings{10.1109/ASONAM49781.2020.9381294,
author = {Alam, Firoj and Ofli, Ferda and Imran, Muhammad and Alam, Tanvirul and Qazi, Umair},
title = {Deep learning benchmarks and datasets for social media image classification for disaster response},
year = {2021},
isbn = {9781728110561},
publisher = {IEEE Press},
url = {https://doi.org/10.1109/ASONAM49781.2020.9381294},
doi = {10.1109/ASONAM49781.2020.9381294},
booktitle = {Proceedings of the 12th IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining},
pages = {151–158},
numpages = {8},
keywords = {benchmarking, crisis computing, deep learning, disaster image classification, natural disasters, social media},
location = {Virtual Event, Netherlands},
series = {ASONAM '20}
}

@inproceedings{mouzannar2018damage,
title={Damage Identification in Social Media Posts using Multimodal Deep Learning.},
author={Mouzannar, Hussein and Rizk, Yara and Awad, Mariette},
booktitle={ISCRAM},
year={2018},
organization={Rochester, NY, USA}
}

@inproceedings{nguyen2017damage,
title={Damage assessment from social media imagery data during disasters},
author={Nguyen, Dat T and Ofli, Ferda and Imran, Muhammad and Mitra, Prasenjit},
booktitle={Proceedings of the 2017 IEEE/ACM international conference on advances in social networks analysis and mining 2017},
pages={569--576},
year={2017}
}
```

# MEDIC：面向灾害图像分类的多任务学习数据集

## 数据

MEDIC是目前规模最大的多任务学习（Multi-Task Learning）灾害相关数据集，为危机图像基准数据集（crisis image benchmark dataset）的扩展版本。该数据集整合了CrisisMMD、AIDR采集的数据以及损伤多模态数据集（Damage Multimodal Dataset, DMD）等多个来源的数据，总计包含71198张图像。

## 目录

- [数据格式与目录](#data-format-and-directories)
- [灾害响应任务](#disaster-response-tasks)
- [下载](#downloads)
- [引用](#citation)
- [使用条款](#terms-of-use)

## 数据格式与目录

### 目录结构

- **data**：主目录，包含以下子目录：
- **aidr_disaster_types/**：存储用于灾害类型任务的、通过AIDR系统采集的图像
- **aidr_info/**：存储用于信息性任务的、通过AIDR系统采集的图像
- **ASONAM17_Damage_Image_Dataset/**：损伤评估数据集
- **crisismmd/**：CrisisMMD数据集
- **multimodal-deep-learning-disaster-response-mouzannar/**：损伤多模态数据集（Damage Multimodal Dataset, DMD）
- **MEDIC_train.tsv、MEDIC_dev.tsv、MEDIC_test.tsv**：分别为训练、开发与测试集文件，采用特定格式
- **LICENSE_CC_BY_NC_SA_4.0.txt**：许可证信息文件
- **terms-of-use.txt**：使用条款文件

### 文件格式

- **image_id**：对应Twitter推文ID或对应来源的图像ID
- **event_name**：事件名称或数据源名称
- **image_path**：图像的相对路径
- **damage_severity**：损伤程度类别标签
- **informative**：信息性类别标签
- **humanitarian**：人道主义类别标签
- **disaster_types**：灾害类型类别标签

## 灾害响应任务

1. **灾害类型**
- 地震（Earthquake）
- 火灾（Fire）
- 洪水（Flood）
- 飓风（Hurricane）
- 滑坡（Landslide）
- 非灾害（Not disaster）
- 其他灾害（Other disaster）

2. **信息性**
- 信息性（Informative）
- 非信息性（Not informative）

3. **人道主义类别**
- 受影响、受伤或遇难人员（Affected, injured, or dead people）
- 基础设施与公用设施损坏（Infrastructure and utility damage）
- 非人道主义（Not humanitarian）
- 救援、志愿或捐赠行动（Rescue, volunteering, or donation effort）

4. **损伤程度评估**
- 轻微或无损伤（Little or no damage）
- 轻度损伤（Mild damage）
- 严重损伤（Severe damage）

## 下载

- **MEDIC数据集 v1.0版**：[下载](https://crisisnlp.qcri.org/data/medic/MEDIC.tar.gz)（11 GB）
- **代码**：[GitHub仓库](https://github.com/firojalam/medic)

### 许可证

MEDIC数据集采用CC BY-NC-SA 4.0许可证发布，允许所有使用者将其用于非商业研究用途，详情可访问：https://creativecommons.org/licenses/by-nc/4.0/。相关许可证信息请参见LICENSE_CC_BY_NC_SA_4.0.txt文件。

## 引用

若您在研究工作中使用该数据集，请引用以下文献：

1. Firoj Alam, Tanvirul Alam, Md. Arid Hasan, Abul Hasnat, Muhammad Imran, Ferda Ofli. *MEDIC: A Multi-Task Learning Dataset for Disaster Image Classification*. Neural Computing and Applications, 35(3):2609–2632, 2023. [论文链接](https://link.springer.com/content/pdf/10.1007/s00521-022-07717-0.pdf) [Arxiv预印本](https://arxiv.org/pdf/2108.12828)
2. Firoj Alam, Ferda Ofli, Muhammad Imran, Tanvirul Alam, Umair Qazi. *Deep Learning Benchmarks and Datasets for Social Media Image Classification for Disaster Response*. 2020 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), 2020.
3. Firoj Alam, Ferda Ofli, and Muhammad Imran. *CrisisMMD: Multimodal Twitter Datasets from Natural Disasters*. Proceedings of the 12th International AAAI Conference on Web and Social Media (ICWSM), 2018, Stanford, California, USA.
4. Hussein Mozannar, Yara Rizk, and Mariette Awad. *Damage Identification in Social Media Posts using Multimodal Deep Learning*. Proc. of ISCRAM, May 2018, pp. 529–543.
5. Dat Tien Nguyen, Ferda Ofli, Muhammad Imran, and Prasenjit Mitra. *Damage assessment from social media imagery data during disasters*. Proc. of ASONAM, pages 1–8, Aug 2017.

@article{alam2022medic,
title={{MEDIC}: A Multi-Task Learning Dataset for Disaster Image Classification},
author={Firoj Alam and Tanvirul Alam and Md. Arid Hasan and Abul Hasnat and Muhammad Imran and Ferda Ofli},
Keywords = {Multi-task Learning, Social media images, Image Classification, Natural disasters, Crisis Informatics, Deep learning, Dataset},
journal={Neural Computing and Applications},
volume={35},
issue={3},
pages={2609--2632},
year={2023},
publisher={Springer}
}

@InProceedings{crisismmd2018icwsm,
author = {Alam, Firoj and Ofli, Ferda and Imran, Muhammad},
title = {{CrisisMMD}: Multimodal Twitter Datasets from Natural Disasters},
booktitle = {Proceedings of the 12th International AAAI Conference on Web and Social Media (ICWSM)},
year = {2018},
month = {June},
date = {23-28},
location = {USA}
}

@inproceedings{10.1109/ASONAM49781.2020.9381294,
author = {Alam, Firoj and Ofli, Ferda and Imran, Muhammad and Alam, Tanvirul and Qazi, Umair},
title = {Deep learning benchmarks and datasets for social media image classification for disaster response},
year = {2021},
isbn = {9781728110561},
publisher = {IEEE Press},
url = {https://doi.org/10.1109/ASONAM49781.2020.9381294},
doi = {10.1109/ASONAM49781.2020.9381294},
booktitle = {Proceedings of the 12th IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining},
pages = {151–158},
numpages = {8},
keywords = {benchmarking, crisis computing, deep learning, disaster image classification, natural disasters, social media},
location = {Virtual Event, Netherlands},
series = {ASONAM '20}
}

@inproceedings{mouzannar2018damage,
title={Damage Identification in Social Media Posts using Multimodal Deep Learning.},
author={Mouzannar, Hussein and Rizk, Yara and Awad, Mariette},
booktitle={ISCRAM},
year={2018},
organization={Rochester, NY, USA}
}

@inproceedings{nguyen2017damage,
title={Damage assessment from social media imagery data during disasters},
author={Nguyen, Dat T and Ofli, Ferda and Imran, Muhammad and Mitra, Prasenjit},
booktitle={Proceedings of the 2017 IEEE/ACM international conference on advances in social networks analysis and mining 2017},
pages={569--576},
year={2017}
}

搜集汇总

背景与挑战

背景概述

MEDIC是一个用于灾害图像分类的多任务学习数据集，包含71,198张图像，支持灾害类型、信息性、人道主义分类和损害严重性评估四个任务。该数据集整合了多个来源的数据，是目前最大的灾害相关多任务学习数据集，适用于非商业研究目的。