合成数据驱动的无人机检测与分类多任务框架

doi:10.1007/s12204-026-2903-3

摘要/Abstract

摘要： 随着无人机技术的快速发展，对智能检测、识别及性能参数推断技术的需求日益迫切。然而，现有的无人机数据集生成方法面临诸多严峻挑战，包括高昂的采集成本、劳动密集型标注以及数据稀缺。目前也缺乏能够从单张图像直接预测功能性能的方法。本文提出了一个基于合成数据的多任务无人机检测分类网络。构建了一个基于Unreal Engine 5的自适应数据集生成系统，结合无人机尺寸分类、自适应距离调整算法以及增强的 3D 到 2D 坐标变换技术，以实现高质量的合成数据自动化生成。设计了一个多任务学习网络，集成视觉信息、距离信息及不确定性量化模块，并支持无人机检测、参数预测、功能分类和尺寸预测的功能。实验结果表明：合成数据与真实数据之间具有高度的相似性（Fréchet初始距离为21.05），且参数预测在最大速度和高度方面的平均绝对百分比误差分别达到 29.1% 和 27.0%，军事-民用分类准确率达到62.5%。该方法为智能无人机分析提供了一种低成本、高效的解决方案，具有相当大的应用价值。

关键词: 无人机, 合成数据集生成, 多任务学习, 参数预测, 不确定性量化

Abstract: With the rapid development of unmanned aerial vehicle (UAV) technology, there is an increasingly urgent demand for intelligent detection, identification, and performance parameter inference techniques. However, existing UAV datasets face severe challenges including high acquisition costs, labor-intensive annotation, and data scarcity, and lack methods for directly predicting functional performance from single images. This paper proposes a UAV analysis framework based on synthetic data generation and multi-task deep learning. We construct an adaptive dataset generation system based on Unreal Engine 5, incorporating UAV size classification, adaptive distance adjustment algorithms, and enhanced 3D-to-2D coordinate transformation techniques for automated high-quality synthetic data generation. We design a multi-task collaborative learning network integrating visual information, distance information, and uncertainty quantification modules, supporting UAV detection, parameter prediction, functional classification, and size prediction. Experimental results demonstrate high similarity between synthetic and real data (Fr´echet inception distance is 21.05), with parameter prediction achieving mean absolute percentage errors of 29.1% and 27.0% for maximum speed and altitude respectively, and military-civilian classification accuracy reaching 62.5%. This method provides a low-cost, efficient solution for intelligent UAV analysis with considerable practical value.

中图分类号:

TP29

. 合成数据驱动的无人机检测与分类多任务框架[J]. J Shanghai Jiaotong Univ Sci, 2026, 31(1): 209-220.

Zhang Jingkai, Li Xinde, Wei Wangzichao, Wang Ziyao, Ma Ke. Synthetic Data-Driven Multi-Task Framework for UAV Detection and Classification[J]. J Shanghai Jiaotong Univ Sci, 2026, 31(1): 209-220.

参考文献

[1] Bikov T, Mihaylov G, Iliev T, et al. Drone surveillance in the modern agriculture [C]//2022 8th International Conference on Energy Efficiency and Agricultural Engineering. Ruse: IEEE, 2022: 1-4.
[2] Qubaa A R, Thannoun R G, Mohammed R M. UAVs/drones for photogrammetry and remote sensing: Nineveh archaeological region as a case study [J]. World Journal of Advanced Research and Reviews, 2022, 14(3): 358-368.
[3] Zhang H H, Tian T, Feng O G, et al. Research on public air route network planning of urban low-altitude logistics unmanned aerial vehicles [J]. Sustainability, 2023, 15(15): 12021.
[4] Wang W, Cheng N, Liu Y L, et al. Content delivery analysis in cellular networks with aerial caching and mmWAVE backhaul [J]. IEEE Transactions on Vehicular Technology, 2021, 70(5): 4809-4822.
[5] Yu N, Mao S, Zhou C, et al. DroneRFa: A large-scale dataset of drone radio frequency signals for detecting low-altitude drones [J]. Journal of Electronics & Information Technology, 2024, 46(4): 1147-1156 (in Chinese).
[6] Perrusquía A, Guo W S, Fraser B, et al. Uncovering drone intentions using control physics informed machine learning [J]. Communications Engineering, 2024, 3: 36.
[7] Khan M A, Menouar H, Eldeeb A, et al. On the detection of unauthorized drones: Techniques and future perspectives: A review [J]. IEEE Sensors Journal, 2022, 22(12): 11439-11455.
[8] Shi X F, Yang C Q, Xie W G, et al. Anti-drone system with multiple surveillance technologies: Architecture, implementation, and challenges [J]. IEEE Communications Magazine, 2018, 56(4): 68-74.
[9] Rahman M H, Sejan M A S, Aziz M A, et al. A comprehensive survey of unmanned aerial vehicles detection and classification using machine learning approach: Challenges, solutions, and future directions [J]. Remote Sensing, 2024, 16(5): 879.
[10] Phung K P, Lu T H, Nguyen T T, et al. Multi-model deep learning drone detection and tracking in complex background conditions [C]//2021 International Conference on Advanced Technologies for Communications. Ho Chi Minh City: IEEE, 2021: 189-194.
[11] Wang B S, Li Q, Mao Q C, et al. A survey on vision-based anti unmanned aerial vehicles methods [J]. Drones, 2024, 8(9): 518.
[12] Unlu E, Zenou E, Riviere N, et al. Deep learning-based strategies for the detection and tracking of drones using several cameras [J]. IPSJ Transactions on Computer Vision and Applications, 2019, 11(1): 7.
[13] Zheng Y, Chen Z, Lv D L, et al. Air-to-air visual detection of micro-UAVs: An experimental evaluation of deep learning [J]. IEEE Robotics and Automation Letters, 2021, 6(2): 1020-1027.
[14] Wisniewski M, Rana Z A, Petrunin I, et al. Drone detection using deep neural networks trained on pure synthetic data [PP/OL]. arXiv (2024-11-13). https://arXiv.org/abs/2411.09077.
[15] Rizzoli G, Barbato F, Caligiuri M, et al. SynDrone–multi-modal UAV dataset for urban scenarios [C]//2023 IEEE/CVF International Conference on Computer Vision Workshops. Paris: IEEE, 2023: 2202-2212.
[16] Lenhard T R, Weinmann A, Franke K, et al. SynDroneVision: A synthetic dataset for image-based drone detection [C]//2025 IEEE/CVF Winter Conference on Applications of Computer Vision. Tucson: IEEE, 2025: 7637-7647.
[17] Dieter T R, Weinmann A, Jäger S, et al. Quantifying the simulation–reality gap for deep learning-based drone detection [J]. Electronics, 2023, 12(10): 2197.
[18] Seidaliyeva U, Ilipbayeva L, Taissariyeva K, et al. Advances and challenges in drone detection and classification techniques: A state-of-the-art review [J]. Sensors, 2024, 24(1): 125.
[19] Liu Z Y, An P, Yang Y, et al. Vision-based drone detection in complex environments: A survey [J]. Drones, 2024, 8(11): 643.
[20] Mrabet M, Sliti M, Ben Ammar L. Machine learning algorithms applied for drone detection and classification: Benefits and challenges [J]. Frontiers in Communications and Networks, 2024, 5: 1440727.
[21] Wisniewski M, Rana Z A, Petrunin I. Drone model classification using convolutional neural network trained on synthetic data [J]. Journal of Imaging, 2022, 8(8): 218.
[22] El-Latif E I A, El-Dosuky M. Predicting power consumption of drones using explainable optimized mathematical and machine learning models [J]. The Journal of Supercomputing, 2025, 81(5): 646.
[23] Bińkowski M, Sutherland D J, Arbel M, et al. Demystifying MMD GANs [PP/OL]. V5. arXiv (2021-01-14) [2025-09-13]. https://arXiv.org/abs/1801.01401.