EarthquakeNet

[ICIP 2024] Official repository for EarthquakeNet: A High-Resolution UAV-Based Dataset for Earthquake Damage Assessment
Published in: 2024 IEEE International Conference on Image Processing (ICIP)

Shenlu Jiang^a, Yuxin Bian^b, Yiran Wang^b, Xufeng Li^a, Zhankeng Liu^a, Yi Ren^a, Yunxuan Zhao^a
a School of Computer Science and Engineering, Macau University of Science and Technology, Avenida Wai Long, Taipa, Macau, China.
^b Faculty of Science, The University of Hong Kong, Pokfulam, Hong Kong.

---

📰 News

• [2024.9.30] The dataset is released on Hugging Face.
• [2024.9.27] Our paper is published in IEEE Xplore.
• [2024.6.7] Our manuscript has been accepted for presentation at IEEE ICIP 2024.

---

📝 Overview

EarthquakeNet is a high-resolution UAV semantic segmentation dataset for post-earthquake damage assessment. It was collected using a fixed-wing UAV after the 2013 Lushan Earthquake in Baoxing County, Sichuan Province, China.

The dataset is distributed as EarthquakeNet_v1.0.zip. After extraction, it contains training and validation images with pixel-level semantic segmentation annotations for 9 classes (8 foreground + background).

---

🗂️ Dataset Statistics

Property	Value
Original UAV images	69
Weather conditions	45 cloud-free / 14 light cloud / 10 heavy cloud
Original resolution	5616 × 3744
Annotation type	Pixel-level semantic segmentation
Number of classes	9 (8 foreground + background)

---

Semantic Classes

ID	Class	RGB
0	Background	(0,0,0)
1	Building – No damage	(0,255,0)
2	Building – Slight/moderate damage	(255,255,0)
3	Building – Bad/heavy damage	(255,120,0)
4	Building – Collapsed	(255,0,0)
5	Road – No damage	(150,150,150)
6	Road – Slight/moderate damage	(150,150,255)
7	Road – Heavy damage	(0,150,255)
8	Tent	(255,0,255)

---

📚 Citation

If you use EarthquakeNet in your research, please cite:

@INPROCEEDINGS{10648157,
  author={Jiang, Shenlu and Bian, Yuxin and Wang, Yiran and Li, Xufeng and Liu, Zhankeng and Ren, Yi and Zhao, Yunxuan},
  booktitle={2024 IEEE International Conference on Image Processing (ICIP)}, 
  title={EarthquakeNet: A High-Resolution UAV-Based Dataset for Earthquake Damage Assessment}, 
  year={2024},
  volume={},
  number={},
  pages={55-61},
  abstract={Advancements in computer vision and deep learning have significantly propelled progress in scene understanding, aiding rescue teams in accurately assessing damage after natural disasters. In this paper, we introduce EarthquakeNet, a meticulously curated high-resolution post-earthquake dataset featuring detailed classification and semantic segmentation annotations, designed to enhance comprehensive scene understanding following natural disasters. EarthquakeNet comprises post-disaster images captured using unmanned aerial vehicles (UAVs) from multiple affected areas after an earthquake. The uniqueness of EarthquakeNet lies in providing high-resolution post-disaster imagery, each with exhaustive annotations. Unlike existing datasets that offer annotations for specific scene elements like buildings, EarthquakeNet provides pixel-level annotations for a broader range of categories, including roads, houses, and tents. We also demonstrate the utility of the dataset by implementing state-of-the-art segmentation models on EarthquakeNet, showcasing its value in improving existing methods for natural disaster damage assessment.},
  keywords={Deep learning;Computer vision;Image resolution;Annotations;Disasters;Semantic segmentation;Roads;Post-earthquake assesment;Land use detection;Semantic segmentation},
  doi={10.1109/ICIP51287.2024.10648157},
  ISSN={2381-8549},
  month={Oct},}

---

📄 License

This dataset is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.

---

📧 Contact

For questions and collaboration, please reach out to Yuxin Bian(yxbian@connect.hku.hk).