AIDERv2 (Aerial Image Dataset for Emergency Response Applications)

SUMMARY OF DATASET • This dataset consist of 16,723 aerial images divided into 4 classes. • The dataset contains three commonly occurring natural disasters earthquake/collapsed buildings, flood, wildfire/fire, and a normal class; do not reflect any disaster • The images can be loaded as numpy arrays using Python programming language and then used to train a Convolutional Neural Network to detect natural disasters from aerial images. • The images are resized to 224x224x3 (heighty,width,channel number) when loaded as numpy arrays. • The dataset is an extension of the AIDER dataset (Aerial Image Dataset for Emergency Response Applications). • Additional images were collected by open source databases and extracted images as frames of videos downloaded from YouTube. The table below shows the number of images in each set. Train Validation Test Total Earthquakes 1927 239 239 2405 Floods 4063 505 502 5070 Fire 3509 439 436 4384 Normal 3900 487 477 4864 Total 13399 1670 1654 16723 If you use this dataset please cite the following publications: [1] Shianios, D., Kyrkou, C., Kolios, P.S. (2023). A Benchmark and Investigation of Deep-Learning-Based Techniques for Detecting Natural Disasters in Aerial Images. In: Tsapatsoulis, N., et al. Computer Analysis of Images and Patterns. CAIP 2023. Lecture Notes in Computer Science, vol 14185. Springer, Cham. https://doi.org/10.1007/978-3-031-44240-7_24 Link: https://link.springer.com/chapter/10.1007/978-3-031-44240-7_24 [2] D. Shianios, P. Kolios, C. Kyrkou, "DiRecNetV2: A Transformer-Enhanced Network for Aerial Disaster Recognition", SN Computer Science, 2024 (Accepted to Appear) DATASET FOLDERS FORMAT └───data │ │ │ └───Dataset_Images │ │ └───Train │ │ │ | └───Earthquake │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Flood │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Normal │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Wildfire │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ └───Val │ │ │ | └───Earthquake │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Flood │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Normal │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Wildfire │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ └───Test │ │ │ | └───Earthquake │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Flood │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Normal │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... │ │ │ | └───Wildfire │ │ │ | img (1).jpg │ │ │ | img (2).jpg │ │ │ | ..... DATA SOURCES AND DATA COLLECTION OPEN SOURCE DATABASES └───AIDER https://zenodo.org/record/3888300#.Yuu11nZBxD- Kyrkou, C. and Theocharides, T., 2020. EmergencyNet: Efficient aerial image classification for drone-based emergency monitoring using atrous convolutional feature fusion. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13, pp.1687-1699. └───ERA https://lcmou.github.io/ERA_Dataset/ Mou, L., Hua, Y., Jin, P. and Zhu, X.X., 2020. Era: A data set and deep learning benchmark for event recognition in aerial videos [software and data sets]. IEEE Geoscience and Remote Sensing Magazine, 8(4), pp.125-133. @article{eradataset, title = {{ERA: A dataset and deep learning benchmark for event recognition in aerial videos}}, author = {Mou, L. and Hua, Y. and Jin, P. and Zhu, X. X.}, journal = {IEEE Geoscience and Remote Sensing Magazine}, year = {in press} } └───ISBDA https://drive.google.com/file/d/1kEKJ8kr1aScXz_1El7Mn-Yi0ANducQIW/view Zhu, X., Liang, J. and Hauptmann, A., 2021. Msnet: A multilevel instance segmentation network for natural disaster damage assessment in aerial videos. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 2023-2032). @misc{zhu2020msnet, title={MSNet: A Multilevel Instance Segmentation Network for Natural Disaster Damage Assessment in Aerial Videos}, author={Xiaoyu Zhu and Junwei Liang and Alexander Hauptmann}, year={2020}, eprint={2006.16479}, archivePrefix={arXiv}, primaryClass={cs.CV} } └───Floods 2013 https://github.com/cvjena/eu-flood-dataset Barz, B., Schröter, K., Münch, M., Yang, B., Unger, A., Dransch, D. and Denzler, J., 2019. Enhancing flood impact analysis using interactive retrieval of social media images. arXiv preprint arXiv:1908.03361. @article{barz2019enhancing, title={Enhancing flood impact analysis using interactive retrieval of social media images}, author={Barz, Bj{\"o}rn and Schr{\"o}ter, Kai and M{\"u}nch, Moritz and Yang, Bin and Unger, Andrea and Dransch, Doris and Denzler, Joachim}, journal={arXiv preprint arXiv:1908.03361}, year={2019} } └───Wildfire Research http://wildfire.fesb.hr/index.php?option=com_content&view=article&id=58&Itemid=54 └───PyImages https://drive.google.com/file/d/1NvTyhUsrFbL91E10EPm38IjoCg6E2c6q/view The dataset was curated by PyImageSearch reader, Gautam Kumar. YOUTUBE VIDEOS └───Collapsed Buildings/Earthquakes • https://www.youtube.com/watch?v=TMow3WPcZrQ&t=133s&ab_channel=GORKHALYFOUNDATION • https://www.youtube.com/watch?v=_HT0tYKKjBI&t=47s&ab_channel=Effect.org • https://www.youtube.com/watch?v=rkb3y6K3waU • https://www.youtube.com/watch?v=yir6ArRZY4o&t=109s&ab_channel=UnicefUK • https://www.youtube.com/watch?v=CM9APmIR9Fk&ab_channel=ToonsZilla • https://www.youtube.com/watch?v=tmx2w6drAeU&ab_channel=AssociatedPress • https://www.youtube.com/watch?v=kuSEe8Emwrk&ab_channel=BloombergQuicktake%3ANow • https://www.youtube.com/watch?v=qoFHA3-m5ag&ab_channel=NBCNews • https://www.youtube.com/watch?v=MM3PToqEPhQ&ab_channel=GuardianNews • https://www.youtube.com/watch?v=zB_-TRnGuZE&ab_channel=DISASTERNEWS • https://www.youtube.com/watch?v=TqAMQQOEsBs&ab_channel=WHAS11 • https://www.youtube.com/watch?v=0ixjTt-jmok&ab_channel=EveningStandard • https://www.youtube.com/watch?v=bNGA8Ms3d70&ab_channel=CatersClips • https://www.youtube.com/watch?v=wJ-2d5t23Lg&ab_channel=DailyDose • https://www.youtube.com/watch?v=ewUcI7I6Gf4&ab_channel=NBCNews • https://www.youtube.com/watch?v=Wx1cjOdlMZ4&ab_channel=ABCNews%28Australia%29 • https://www.youtube.com/watch?v=jiMK_sVmbXk&t=12s&ab_channel=NewChinaTV • https://www.youtube.com/watch?v=M9au_9A2YRo&ab_channel=GuardianNews • https://www.youtube.com/watch?v=i6Lh8IXPjso&ab_channel=TheSun • https://www.youtube.com/watch?v=CKwxEr3I4Y8&ab_channel=GuardianNews • https://www.youtube.com/watch?v=hxqzcajBCNg&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=3&ab_channel=WXChasing • https://www.youtube.com/watch?v=2GEeTDuf9mI&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=6&ab_channel=WXChasing • https://www.youtube.com/watch?v=bDOuZWxIyNQ&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=9&ab_channel=WXChasing • https://www.youtube.com/watch?v=vzoSADijLCQ&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=15&ab_channel=WXChasing • https://www.youtube.com/watch?v=ZaL1fldTEAk&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=17&ab_channel=WXChasing • https://www.youtube.com/watch?v=QSV81FdilZE&ab_channel=GlobalNews • https://www.youtube.com/watch?v=KgOk9otW1Bg&ab_channel=EricFeijten └───Floods • https://www.youtube.com/watch?v=DJqgv8Sa5bA&t=317s&ab_channel=7NEWSAustralia • https://www.youtube.com/watch?v=w5FintiCLJU&t=9s&ab_channel=GuardianNews • https://www.youtube.com/watch?v=HjMymNN6Ajc&t=143s&ab_channel=BioLogicTreeServices • https://www.youtube.com/watch?v=Tmba18C94C8&ab_channel=AL.com • https://www.youtube.com/watch?v=Dqvpv4Vg4lk&t=63s&ab_channel=ElevenEleven • https://www.youtube.com/watch?v=N7QGicNtN2A&ab_channel=PKSVideoProductions • https://www.youtube.com/watch?v=8CHagyQG16Q&ab_channel=Stolly-Sven • https://www.youtube.com/watch?v=vjH3zFqdzcE&ab_channel=BenChilders • https://www.youtube.com/watch?v=GFw89UB4fE8&ab_channel=BenChilders • https://www.youtube.com/watch?v=heP3LEJ_NkE&ab_channel=7NEWSAustralia └───Fires • https://www.youtube.com/watch?v=gbM_NPx2GPc&t=201s&ab_channel=WXChasing • https://www.youtube.com/watch?v=M97sJdyeEM4&t=72s&ab_channel=Sanuck176 • https://www.youtube.com/watch?v=1Z2K6lDt76M&t=557s&ab_channel=TheRelaxationChannel └───Normal • https://www.youtube.com/watch?v=SyxjsuNHWhM&t=328s&ab_channel=OneManWolfPack • https://www.youtube.com/watch?v=f1PTWsBtrtc&ab_channel=ChernobylPug## 数据集概述 • 本数据集包含16723张航拍图像，共分为4个类别。 • 本数据集涵盖三类常见自然灾害：地震/建筑坍塌、洪涝、野火/火灾，以及一个无灾害的正常类别。 • 可通过Python编程语言将图像加载为NumPy数组，进而用于训练卷积神经网络(Convolutional Neural Network)以实现航拍图像中的自然灾害检测任务。 • 当图像被加载为NumPy数组时，已被统一调整为224×224×3的尺寸（高度、宽度、通道数）。 • 本数据集是AIDER数据集(Aerial Image Dataset for Emergency Response Applications，应急响应应用航拍图像数据集)的扩展版本。 • 额外的图像数据通过开源数据库收集，以及从YouTube下载的视频帧中提取得到。 下表展示了各子集的图像数量： | 类别 | 训练集 | 验证集 | 测试集 | 总计 | |--------------|--------|--------|--------|--------| | 地震 | 1927 | 239 | 239 | 2405 | | 洪涝 | 4063 | 505 | 502 | 5070 | | 火灾（野火） | 3509 | 439 | 436 | 4384 | | 正常类别 | 3900 | 487 | 477 | 4864 | | 总计 | 13399 | 1670 | 1654 | 16723 | 若使用本数据集，请引用以下文献： [1] Shianios D, Kyrkou C, Kolios P S. (2023). A Benchmark and Investigation of Deep-Learning-Based Techniques for Detecting Natural Disasters in Aerial Images. In: Tsapatsoulis, N., et al. Computer Analysis of Images and Patterns. CAIP 2023. Lecture Notes in Computer Science, vol 14185. Springer, Cham. 链接: https://link.springer.com/chapter/10.1007/978-3-031-44240-7_24 [2] D. Shianios, P. Kolios, C. Kyrkou, "DiRecNetV2: A Transformer-Enhanced Network for Aerial Disaster Recognition", SN Computer Science, 2024 (已录用待刊) ## 数据集文件夹格式 └───data │ │ └───Dataset_Images │ │ │ └───Train │ │ │ │ │ └───Earthquake │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ │ └───Flood │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ │ └───Normal │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ │ └───Wildfire │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ └───Val │ │ │ │ │ └───Earthquake │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ │ └───Flood │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ │ └───Normal │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ │ └───Wildfire │ │ │ img (1).jpg │ │ │ img (2).jpg │ │ │ ...... │ └───Test │ │ │ └───Earthquake │ │ img (1).jpg │ │ img (2).jpg │ │ ...... │ └───Flood │ │ img (1).jpg │ │ img (2).jpg │ │ ...... │ └───Normal │ │ img (1).jpg │ │ img (2).jpg │ │ ...... │ └───Wildfire │ img (1).jpg │ img (2).jpg │ ...... ## 数据集来源与采集 ### 开源数据库 #### AIDER 数据集链接：https://zenodo.org/record/3888300#.Yuu11nZBxD- 相关文献：Kyrkou, C. and Theocharides, T., 2020. EmergencyNet: Efficient aerial image classification for drone-based emergency monitoring using atrous convolutional feature fusion. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13, pp.1687-1699. #### ERA 数据集链接：https://lcmou.github.io/ERA_Dataset/ 相关文献：Mou, L., Hua, Y., Jin, P. and Zhu, X.X., 2020. Era: A data set and deep learning benchmark for event recognition in aerial videos [software and data sets]. IEEE Geoscience and Remote Sensing Magazine, 8(4), pp.125-133. 附BibTeX引用： @article{eradataset, title = {{ERA: A dataset and deep learning benchmark for event recognition in aerial videos}}, author = {Mou, L. and Hua, Y. and Jin, P. and Zhu, X. X.}, journal = {IEEE Geoscience and Remote Sensing Magazine}, year = {in press} } #### ISBDA 数据集链接：https://drive.google.com/file/d/1kEKJ8kr1aScXz_1El7Mn-Yi0ANducQIW/view 相关文献：Zhu, X., Liang, J. and Hauptmann, A., 2021. Msnet: A multilevel instance segmentation network for natural disaster damage assessment in aerial videos. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 2023-2032). 附BibTeX引用： @misc{zhu2020msnet, title={MSNet: A Multilevel Instance Segmentation Network for Natural Disaster Damage Assessment in Aerial Videos}, author={Xiaoyu Zhu and Junwei Liang and Alexander Hauptmann}, year={2020}, eprint={2006.16479}, archivePrefix={arXiv}, primaryClass={cs.CV} } #### 2013年洪涝数据集 数据集链接：https://github.com/cvjena/eu-flood-dataset 相关文献：Barz, B., Schröter, K., Münch, M., Yang, B., Unger, A., Dransch, D. and Denzler, J., 2019. Enhancing flood impact analysis using interactive retrieval of social media images. arXiv preprint arXiv:1908.03361. 附BibTeX引用： @article{barz2019enhancing, title={Enhancing flood impact analysis using interactive retrieval of social media images}, author={Barz, Björn and Schrötter, Kai and Münch, Moritz and Yang, Bin and Unger, Andrea and Dransch, Doris and Denzler, Joachim}, journal={arXiv preprint arXiv:1908.03361}, year={2019} } #### 野火研究数据集 数据集链接：http://wildfire.fesb.hr/index.php?option=com_content&view=article&id=58&Itemid=54 #### PyImages 数据集链接：https://drive.google.com/file/d/1NvTyhUsrFbL91E10EPm38IjoCg6E2c6q/view 该数据集由PyImageSearch读者Gautam Kumar整理。 ### YouTube视频来源 #### 坍塌建筑/地震类图像 以下为采集所用的YouTube视频链接： • https://www.youtube.com/watch?v=TMow3WPcZrQ&t=133s&ab_channel=GORKHALYFOUNDATION • https://www.youtube.com/watch?v=_HT0tYKKjBI&t=47s&ab_channel=Effect.org • https://www.youtube.com/watch?v=rkb3y6K3waU • https://www.youtube.com/watch?v=yir6ArRZY4o&t=109s&ab_channel=UnicefUK • https://www.youtube.com/watch?v=CM9APmIR9Fk&ab_channel=ToonsZilla • https://www.youtube.com/watch?v=tmx2wdrAeU&ab_channel=AssociatedPress • https://www.youtube.com/watch?v=kuSE8Emwrk&ab_channel=BloombergQuicktake%3ANow • https://www.youtube.com/watch?v=qoFHA3-m5ag&ab_channel=NBCNews • https://www.youtube.com/watch?v=MM3PToqEPhQ&ab_channel=GuardianNews • https://www.youtube.com/watch?v=zB_-TRnGuZE&ab_channel=DISASTERNEWS • https://www.youtube.com/watch?v=TqAMQQOEsBs&ab_channel=WHAS11 • https://www.youtube.com/watch?v=0ixjTt-jmok&ab_channel=EveningStandard • https://www.youtube.com/watch?v=bNGA8Ms3d70&ab_channel=CatersClips • https://www.youtube.com/watch?v=wJ-2d5t23Lg&ab_channel=DailyDose • https://www.youtube.com/watch?v=ewUcI7I6Gf4&ab_channel=NBCNews • https://www.youtube.com/watch?v=Wx1cjOdlMZ4&ab_channel=ABCNews%28Australia%29 • https://www.youtube.com/watch?v=jiMK_sVmbXk&t=12s&ab_channel=NewChinaTV • https://www.youtube.com/watch?v=M9au_9A2YRo&ab_channel=GuardianNews • https://www.youtube.com/watch?v=i6Lh8IXPjso&ab_channel=TheSun • https://www.youtube.com/watch?v=CKwxEr3I4Y8&ab_channel=GuardianNews • https://www.youtube.com/watch?v=hxqzcajBCNg&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=3&ab_channel=WXChasing • https://www.youtube.com/watch?v=2GEeTDuf9mI&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=6&ab_channel=WXChasing • https://www.youtube.com/watch?v=bDOuZWxIyNQ&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=9&ab_channel=WXChasing • https://www.youtube.com/watch?v=vzoSADijLCQ&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=15&ab_channel=WXChasing • https://www.youtube.com/watch?v=ZaLfldTEAk&list=RDCMUCD3KREyo3IqCLBC-4khGgIw&index=17&ab_channel=WXChasing • https://www.youtube.com/watch?v=QSV81FdilZE&ab_channel=GlobalNews • https://www.youtube.com/watch?v=KgOk9otW1Bg&ab_channel=EricFeijten #### 洪涝类图像 以下为采集所用的YouTube视频链接： • https://www.youtube.com/watch?v=DJqgv8Sa5bA&t=317s&ab_channel=7NEWSAustralia • https://www.youtube.com/watch?v=w5FintiCLJU&t=9s&ab_channel=GuardianNews • https://www.youtube.com/watch?v=HjMymNN6Ajc&t=143s&ab_channel=BioLogicTreeServices • https://www.youtube.com/watch?v=Tmba18C94C8&ab_channel=AL.com • https://www.youtube.com/watch?v=Dqvpv4Vg4lk&t=63s&ab_channel=ElevenEleven • https://www.youtube.com/watch?v=N7QGicNtN2A&ab_channel=PKSVideoProductions • https://www.youtube.com/watch?v=8CHagyQG16Q&ab_channel=Stolly-Sven • https://www.youtube.com/watch?v=vjH3zFqdzcE&ab_channel=BenChilders • https://www.youtube.com/watch?v=GFw89UB4fE8&ab_channel=BenChilders • https://www.youtube.com/watch?v=heP3LEJ_NkE&ab_channel=7NEWSAustralia #### 火灾类图像 以下为采集所用的YouTube视频链接： • https://www.youtube.com/watch?v=gbM_NPx2GPc&t=201s&ab_channel=WXChasing • https://www.youtube.com/watch?v=M97sJdyeEM4&t=72s&ab_channel=Sanuck176 • https://www.youtube.com/watch?v=1Z2K6lDt76M&t=557s&ab_channel=TheRelaxationChannel #### 正常类别图像 以下为采集所用的YouTube视频链接： • https://www.youtube.com/watch?v=SyxjsuNHWhM&t=328s&ab_channel=OneManWolfPack • https://www.youtube.com/watch?v=f1PTWsBtrtc&ab_channel=ChernobylPug