Long-Duration Drone Tracking Dataset

Associated Paper: Detector-Augmented SAMURAI for Long-Duration Drone Tracking

Published in the Proceedings of the 2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops!

The Long-Duration Drone Tracking Dataset is part of the paper "Detector-Augmented SAMURAI for Long-Duration Drone Tracking" by Tamara R. Lenhard, Andreas Weinmann, Hichem Snoussi, Tobias Koch. This work was presented in the 2026 IEEE/CVF WACV Workshop on Real-World Surveillance: Applications and Challenges (RWS) and is published in the Proceedings of the 2026 IEEE/CVF WACV Workshops.

Final Version: here

(ArXiv Preprint: here)

Dataset Details

The Long-Duration Drone Tracking Dataset consists of two subsets, R1 and R2, each containing two long-duration sequences. Every sequence is recorded at a resolution of 2040 × 1086 pixels using a ground-mounted Basler acA200-165c camera system equipped with two different lenses (R1: 8 mm lens, R2: 25 mm lens). The recordings were captured in an urban environment characterized by medium-density vegetation and medium-height buildings.

Annotation Format: Annotations (bounding boxes) are provided via text files according to the YOLO standard format

<object-class> <x> <y> <width> <height>

Here, <x> and <y> represent the normalized coordinates of the bounding box center, while <width> and <height> denote the normalized bounding box wisth and height. In SynDroneVision, <object-class> is always 0, indicating the drone class.

Download

Image data and annotations are available via the following links:

Subset Name
Sequence
Size 
Download Link

R1
POS3
7.1 GB
Download R1_POS3

R1
POS7
4.7 GB
Download R1_POS7

R2
POS3
1.8 GB
Download R2_POS3

R2
POS7
5.4 GB
Download R2_POS7

Example images of each sequence are shown below!

Citation

If you find the Long-Duration Drone Tracking Dataset helpful in your research, we kindly ask that you cite the associated paper. Below is the citation in BibTeX format for your convenience:

BibTeX:

@InProceedings{Lenhard_2026_WACV, author = {Lenhard, Tamara R. and Weinmann, Andreas and Snoussi, Hichem and Koch, Tobias}, title = {Detector-Augmented SAMURAI for Long-Duration Drone Tracking}, booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops}, month = {March}, year = {2026}, pages = {75-84} }

关联论文：《面向长时无人机追踪的检测器增强型SAMURAI》
本研究发表于2026年IEEE/CVF冬季计算机视觉应用会议（WACV）研讨会论文集！

长时无人机追踪数据集（Long-Duration Drone Tracking Dataset）由Tamara R. Lenhard、Andreas Weinmann、Hichem Snoussi、Tobias Koch共同撰写的同名论文《面向长时无人机追踪的检测器增强型SAMURAI》所收录。本工作展示于2026年IEEE/CVF WACV真实世界监控：应用与挑战（RWS）研讨会，并收录于2026年IEEE/CVF WACV研讨会论文集。

最终版本：[此处链接]
（ArXiv预印本：[此处链接]）

数据集详情
长时无人机追踪数据集包含R1与R2两个子集，每个子集均包含两段长时序列。所有序列均采用搭载两种不同镜头的地面部署Basler acA200-165c相机系统录制，分辨率为2040×1086像素（R1子集使用8mm镜头，R2子集使用25mm镜头）。录制场景为兼具中等密度植被与中等高度建筑的城市环境。

标注格式：标注（边界框）以YOLO标准格式的文本文件形式提供，格式如下：
<目标类别> <x> <y> <宽度> <高度>
其中<x>与<y>代表边界框中心的归一化坐标，<宽度>与<高度>表示归一化后的边界框宽高。在SynDroneVision数据集中，<目标类别>恒为0，代表无人机类别。

下载
图像数据与标注可通过以下链接获取：

| 子集名称 | 序列编号 | 大小 | 下载链接 |
|--------|--------|-------|--------|
| R1 | POS3 | 7.1 GB | 下载R1_POS3 |
| R1 | POS7 | 4.7 GB | 下载R1_POS7 |
| R2 | POS3 | 1.8 GB | 下载R2_POS3 |
| R2 | POS7 | 5.4 GB | 下载R2_POS7 |

各序列的示例图像如下所示！

引用
若您在研究中使用长时无人机追踪数据集，请引用本关联论文。为方便您使用，以下为BibTeX格式的引用信息：

BibTeX：
@InProceedings{Lenhard_2026_WACV, author = {Lenhard, Tamara R. and Weinmann, Andreas and Snoussi, Hichem and Koch, Tobias}, title = {Detector-Augmented SAMURAI for Long-Duration Drone Tracking}, booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops}, month = {March}, year = {2026}, pages = {75-84} }