AIResQ: High-resolution thermal infrared dataset for airborne person detection

AIResQ is a dataset of thermal IR images with a resolution up to 2048x1536 pixel. The dataset comprises 9,788 thermal images with 17,550 labeled persons in SAR related scenarios. The whole dataset includes images from a full annual cycle, on different times of day, within several types of terrain and different weather patterns. The number of individuals visible in each image varies between zero and fifteen. Terms of Use: This dataset is provided for civilian and non-military research purposes only. We explicitly distance ourselves from any non-civilian use of this data. Its intended scope is strictly limited to peaceful and civilian progress. Citation: A research paper was submitted to Scientific Data. If you use the AIResQ dataset in your research or project, please cite the corresponding article. The reference will be updated on acceptance. Büttner et al, "High-resolution thermal infrared dataset for airborne person detection in SAR missions", submitted to Scientific Data, 2026. @article{AIResQ_2026, author = {Büttner, Johannes and Führer, Kilian and Fritz, Andreas and Zender, Jonathan and Pinzer, Bernd R.}, title = {High-resolution thermal infrared dataset for airborne person detection in SAR missions}, journal = {submitted to Scientific Data}, year = {2026}, doi = {10.5281/zenodo.19677785}, url = {https://zenodo.org/records/19677785}}AIResQ是一款高分辨率热红外（thermal IR）图像数据集，最大分辨率可达2048×1536像素。该数据集共包含9788张热红外图像，其中标注有17550个人物目标，所有场景均与合成孔径雷达（Synthetic Aperture Radar, SAR）任务相关。本数据集覆盖完整的年度周期，涵盖不同时段、多种地形类型以及多样天气条件下采集的图像；单张图像中可见的人物数量区间为0至15。 使用条款： 本数据集仅可用于民用及非军事科研用途，严禁将其用于非民用场景。 引用说明： 一篇相关研究论文已提交至Scientific Data期刊。若您在研究或项目中使用AIResQ数据集，请引用该对应论文，最终参考文献信息将在论文录用后更新。 Büttner 等人，《用于合成孔径雷达任务中机载人员检测的高分辨率热红外数据集》，已提交至 Scientific Data，2026年。 @article{AIResQ_2026, author = {Büttner, Johannes and Führer, Kilian and Fritz, Andreas and Zender, Jonathan and Pinzer, Bernd R.}, title = {High-resolution thermal infrared dataset for airborne person detection in SAR missions}, journal = {submitted to Scientific Data}, year = {2026}, doi = {10.5281/zenodo.19677785}, url = {https://zenodo.org/records/19677785}}