基于STOD的无人机舰船图像微小目标检测方法

doi:10.12305/j.issn.1001-506X.2025.11.05

摘要/Abstract

摘要：

针对无人机航拍舰船图像小目标占比高、检测难度大的问题，提出一种舰船微小目标检测（ship tiny object detection，STOD）算法。算法立足于图像特性、舰船特征并结合目标特定场景，重点考虑无人机对STOD过程中的难点问题，针对性地设计了算法模型和网络框架。首先，采用轻量化的单阶段无锚框网络作为主干，然后引入全局注意力机制，强化模型对微小目标的学习能力，随后设计局部跨阶段金字塔聚合网络并强化检测头，提升网络对目标特征的提取能力和检测精度，最后在回归分支中采用基于向量的分布焦点损失，进一步强化对微小目标的检测能力。实验验证和对比分析表明，所提方法对舰船微小目标的检测性能相比于基线模型平均精度整体提升了2.6%，每秒处理帧数可达52.2。对比其他小目标算法检测精度高，实时性好。

关键词: 无人机, 舰船图像, 小目标检测, 深度学习, 全局注意力机制

Abstract:

A ship tiny object detection（STOD ）is proposed to address the problem of high proportion of small targets and difficult detection in unmanned aerial vehicle ship images. The algorithm is based on image characteristics, ship features, and combined with target specific scenes, focusing on the difficult problems in unmanned aerial vehicle STOD process. A targeted algorithm model and network framework are designed. Firstly, a lightweight single-stage anchor free network is adopted as the backbone, followed by the introduction of a global attention mechanism to enhance the model’s learning ability for tiny objects. Then, cross-stage partial pyramid aggregation network is designed and the detection head is strengthened to improve the network’s ability to extract target features and detection accuracy. Finally, a vector based distribution focus loss is used in the regression branch to further enhance the detection ability for tiny objects. Experimental verification and comparative analysis indicate that the proposed method improves the detection performance of tiny objects on ships by 2.6% compared to the baseline model average precision, and the frame per second can reach 52.2. Compared with other small target algorithms, it has high detection accuracy and good real-time performance.

Key words: unmanned aerial vehicle （UAV）, ship image, tiny objects detection, deep learning, global attention mechanism

中图分类号:

TP 391.4

姜杰, 闫文君, 凌青, 张立民. 基于STOD的无人机舰船图像微小目标检测方法[J]. 系统工程与电子技术, 2025, 47(11): 3559-3567.

Jie JIANG, Wenjun YAN, Qing LING, Limin ZHANG. Tiny objects detection method for unmanned aerial vehicle ship images based on STOD[J]. Systems Engineering and Electronics, 2025, 47(11): 3559-3567.

图/表 15

图1

图2

图3

图4

图5

图6

表1

图7

图8

表2

图9

表3

表4

图10

图11

参考文献 31

1	ZHANG H, CAO C H, XU L W, et al. A UAV detection algorithm based on an artificial neural network[J]. IEEE Access, 2018, 6, 24720- 24728. doi: 10.1109/ACCESS.2018.2831911
2	REKAVANDI A M, RASHIDI S, BOUSSAID F, et al. Transformers in small object detection: a benchmark and survey of state-of-the-art[J]. IEEE Trans. on Image Processing, 2021, 30, 5017- 5031. doi: 10.1109/TIP.2021.3077139
3	HASSAN S A, RAHIM T, SHIN S Y. Real-time UAV detection based on deep learning network[C]//Proc. of the International Conference on Information and Communication Technology Convergence, 2019.
4	OPROMALLA R, FASANO G, ACCARDO D. A vision-based approach to UAV detection and tracking in cooperative applications[J]. Sensors, 2018, 18 (10): 3391. doi: 10.3390/s18103391
5	KOU R, WANG C P, PENG Z M, et al. Infrared small target segmentation networks: a survey[J]. Pattern Recognition, 2023, 143, 109788. doi: 10.1016/j.patcog.2023.109788
6	BAI X Z, CHEN Z G, ZHANG Y, et al. Infrared ship target segmentation based on spatial information improved FCM[J]. IEEE Trans. on Cybernetics, 2015, 46 (12): 3259- 3271.
7	LIU Z Y, ZHOU F G, CHEN X W, et al. Iterative infrared ship target segmentation based on multiple features[J]. Pattern Recognition, 2014, 47 (9): 2839- 2852. doi: 10.1016/j.patcog.2014.03.005
8	BAI X Z, LIU M M, WANG T, et al. Feature based fuzzy inference system for segmentation of low-contrast infrared ship images[J]. Applied Soft Computing, 2016, 46, 128- 142. doi: 10.1016/j.asoc.2016.05.004
9	SONG Z Z, YANG J W, ZHANG D F, et al. Semi-supervised dim and small infrared ship detection network based on haar wavelet[J]. IEEE Access, 2021, 9, 29686- 29695. doi: 10.1109/ACCESS.2021.3058526
10	REN Y M, YANG J, GUO Z Q, et al. Ship classification based on attention mechanism and multi-scale convolutional neural network for visible and infrared images[J]. Electronics, 2020, 9 (12): 2022. doi: 10.3390/electronics9122022
11	HAN J H, MA Y, ZHOU B, et al. On human visual system[J]. IEEE Geoscience and Remote Sensing Letters, 2014, 11 (12): 2168- 2172. doi: 10.1109/LGRS.2014.2323236
12	WANG K Y, DU S Y, LIU C X, et al. Interior attention-aware network for infrared small target detection[J]. IEEE Trans. on Geoscience and Remote Sensing, 2022, 60: 5002013.
13	ZHANG M J, ZHANG R, YANG Y X, et al. ISNet: shape matters for infrared small target detection[C]//Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022: 877−886.
14	ZHAO B, WANG C P, FU Q, et al. A novel pattern for infrared small target detection with generative adversarial network[J]. IEEE Trans. on Geoscience and Remote Sensing, 2020, 59 (5): 4481- 4492.
15	HOU Q Y, WANG Z P, TAN F J, et al. RISTDNet: robust infrared small target detection network[J]. IEEE Geoscience and Remote Sensing Letters, 2021, 19, 7000805.
16	DAI Y Y, WU Y Q, ZHOU F, et al. Attentional local contrast networks for infrared small target detection[J]. IEEE Trans. on Geoscience and Remote Sensing, 2021, 59 (11): 9813- 9824. doi: 10.1109/TGRS.2020.3044958
17	THECKEDATH D, SEDAMKAR R R. Detecting affect states using VGG16, ResNet50 and SE-ResNet50 networks[J]. SN Computer Science, 2020, 1: 79.
18	WANG C Y, LIAO H, WU Y H, et al. CSPNet: a new backbone that can enhance learning capability of CNN[C]//Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. 2020: 390−391.
19	XU W J, LONG C J, WANG R S, et al. DRB-GAN: a dynamic resblock generative adversarial network for artistic style transfer[C]//Proc. of the IEEE/CVF International Conference on Computer Vision, 2021: 6383−6392.
20	XU G Y, XU Y, ZHANG S C, et al. SFRNet: feature extraction-fusion steganalysis network based on squeeze-and-excitation block and RepVgg block[J]. Security and Communication Networks, 2021, 1, 3676720.
21	YANG J F, FU X Y, HU Y W, et al. PanNet: a deep network architecture for pan-sharpening[C]//Proc. of the IEEE International Conference on Computer Vision, 2017: 5449−5457.
22	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[EB/OL]. [2025-02-04]. https: //arxiv.org/abs/1706.03762.
23	GONG Y Q, YU X H, DING Y, et al. Effective fusion factor in FPN for tiny object detection[C]//Proc. of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2021: 1160−1168.
24	FENG C, ZHONG Y, GAO Y, et al. Tood: task-aligned one-stage object detection[C]//Proc. of the IEEE/CVF International Conference on Computer Vision, 2021: 3490−3499.
25	姜杰, 张立民, 刘凯, 等. 基于改进PP-YOLOE和ByteTrack算法的红外船舶目标检测跟踪方法[J]. 兵器装备工程学报, 2024, 45 (11): 291- 297.
	JIANG J, ZHANG L M, LIU K, et al. Infrared ship target detection and tracking method based on improved PP-YOLOE and Bytetrack algorithms[J]. Journal of Weapon Equipment Engineering, 2024, 45 (11): 291- 297.
26	姜杰, 凌青, 闫文君, 等. 基于MFFDet-R的多源舰船图像融合检测方法[J]. 系统工程与电子技术, 2025, 47 (2): 390- 397.
	JIANG J, LING Q, YAN W J, et al. Multi source ship image fusion detection method based on MFFDet-R[J]. Systems Engineering and Electronics, 2025, 47 (2): 390- 397.
27	MALHOTRA R, SHAKYA A, RANJAN R, et al. Software defect prediction using binary particle swarm optimization with binary cross entropy as the fitness function[J] Journal of Physics: Conference Series, 2021, 1767(1): 012003.
28	YANG X, YANG X J, YANG J R, et al. Learning high-precision bounding box for rotated object detection via kullback-leibler divergence[J]. Advances in Neural Information Processing Systems, 2021, doi: 10.48550/arXiv.2160.01883
29	LI X, WANG W H, WU L J, et al. Generalized focal loss: learning qualified and distributed bounding boxes for dense object detection[J]. Advances in Neural Information Processing Systems, 2020, 33, 21002- 21012.
30	FU A M, ZHANG X L, ZHONG N X, et al. VFL: a verifiable federated learning with privacy-preserving for big data in industrial IoT[J]. IEEE Trans. on Industrial Informatics, 2020, 18 (5): 3316- 3326.
31	姜杰, 张立民, 刘凯, 等. 基于任务对齐学习的红外船舶目标检测方法[J]. 系统工程与电子技术, 2025, 47 (1): 34- 40.
	JIANG J, ZHANG L M, LIU K, et al. Infrared ship target detection method based on task alignment learning[J]. Systems Engineering and Electronics, 2025, 47 (1): 34- 40.

序号	类别	数量
1	中目标	7541
2	小目标	34215
3	微小目标	25359

模块	AP/%	参数量（×10⁶）	运算量/ GFLOPs	时延/ ms	FPS
基线模型	47.4	50.63	110.75	12.9	55.4
增加无锚框	47.1（−0.3）	48.54	108.47	12.1	56.3
改进网络结构	47.5（+0.4）	46.32	104.51	11.2	58.7
+全局注意力模块	48.0（+0.5）	48.42	106.87	12.2	56.1
+CSPPAN	48.7（+0.7）	52.31	110.07	13.1	55.3
ET-head	49.6（+0.9）	53.68	114.78	14.1	53.3
+基于向量DFL	50.1（+0.5）	55.58	115.77	15.3	52.2

参数组合	AP/%	AP₅₀ /%	推理速度/（帧/s）
α=1.0, β=0.5, γ=0.5	33.2	56.7	53.1
α=1.2, β=0.6, γ=0.4	33.8	58.3	52.7
α=0.8, β=1.0, γ=0.3	33.1	55.2	53.6

算法名称	AP	AP₅₀	AP₇₅	AP_T	AP_S	AP_M
扩展FPN	29.13	54.07	27.38	29.64	32.98	77.13
大核卷积与层次特征融合	28.55	56.63	26.68	29.68	33.15	74.51
多尺度特征提取与跨阶段特征融合网络	27.33	49.56	26.55	32.13	36.84	78.01
密集卷积注意力网络	26.46	48.34	25.38	31.45	35.71	76.49
多重注意力轻量化网络	26.35	47.98	25.45	32.09	36.78	76.17
YOLOX	31.28	60.39	29.61	32.15	37.14	77.84
YOLOv8-p2	33.52	62.41	32.16	32.44	37.79	78.11
YOLOv9	33.64	62.56	32.45	32.49	37.83	78.31
YOLOv10	33.84	62.78	32.31	32.61	38.02	78.40
本文算法	34.61	63.25	32.54	32.72	38.16	78.47

[1]	李开明, 张袁鹏, 罗迎, 代肖楠. 弹道导弹雷达目标识别研究进展[J]. 系统工程与电子技术, 2025, 47(9): 2870-2889.
[2]	闻雯, 时晨光, 周建江. 多元威胁环境下无人机集群隐身航迹规划算法[J]. 系统工程与电子技术, 2025, 47(9): 2971-2984.
[3]	魏潇龙, 吴亚荣, 姚登凯, 赵顾颢. 基于深度强化学习的无人机空战机动分层决策算法[J]. 系统工程与电子技术, 2025, 47(9): 2993-3003.
[4]	张国庆, 徐轶晖, 李纪强, 张显库, 邱斌. 基于异步搜寻制导的机/船协同事件触发控制[J]. 系统工程与电子技术, 2025, 47(9): 3058-3065.
[5]	杨大鹏, 龚资浩, 王小也, 郭正玉, 罗德林. 基于多智能体强化学习的无人机协同截击机动决策研究[J]. 系统工程与电子技术, 2025, 47(9): 3076-3085.
[6]	季策, 马相宇, 牟晓宇, 赵家毅. TS-GRU-VTA：基于深度学习的车辆信道估计方案[J]. 系统工程与电子技术, 2025, 47(9): 3093-3098.
[7]	付卫红, 彭文洪, 刘乃安. 混合注意力优化的SAR图像小目标检测方法[J]. 系统工程与电子技术, 2025, 47(8): 2519-2526.
[8]	张欣悦, 吴晓莉, 王名珺, 晏彪, 武愈涵. 有/无人机协同操作界面的最佳交互方式评估[J]. 系统工程与电子技术, 2025, 47(8): 2600-2611.
[9]	李延通, 李子璠, 周姗姗, 张闯. 无人机光伏电站巡检双目标选址-路径问题研究[J]. 系统工程与电子技术, 2025, 47(8): 2612-2621.
[10]	羊钊, 胡锦标, 王艳, 齐洪彪. 考虑异巢起降的无人机山地巡检覆盖路径规划[J]. 系统工程与电子技术, 2025, 47(8): 2622-2631.
[11]	闫小伟, 凌冲, 石胜斌. 地表未爆子弹药快速检测系统设计与实现[J]. 系统工程与电子技术, 2025, 47(8): 2639-2645.
[12]	张晓璐, 陈亚洲, 赵敏. 无人机数据链带内电磁干扰效应预测模型与验证[J]. 系统工程与电子技术, 2025, 47(8): 2763-2773.
[13]	符小卫, 王辛夷, 乔哲. 基于APIQ算法的多无人机攻防对抗策略[J]. 系统工程与电子技术, 2025, 47(7): 2205-2215.
[14]	郑凯文, 杜承泽, 赵兴芳, 逄晓凡. 融合时空散列的三维RRT^*多编队航路规划[J]. 系统工程与电子技术, 2025, 47(7): 2256-2266.
[15]	吴北苹, 何晶, 党慧莹, 岳地久. 基于作战环的反无人机作战体系贡献率评估[J]. 系统工程与电子技术, 2025, 47(7): 2267-2274.