MCS-RETR：改进RT-DETR的无人机航拍图像目标检测方法

doi:10.7527/S1000-6893.2025.31987

Abstract

Abstract:

To address the challenges in Unmanned Aerial Vehicle （UAV） aerial imagery， such as blurred object edge information， complex background interference， and the low resolution and difficulty in identification of small objects， this paper proposes a novel improved RT-DETR-based object detection method for UAV aerial imagery， named MCS-DETR. Initially， a multi-scale edge information enhancement module is designed within the backbone network. By constructing an edge information enhancement mechanism and integrating it with deep convolutional operations and feature fusion strategies， the module aims to extract feature information at different scales and enhance the model’s perception of image edge information. Subsequently， a convolutional additive token mixer is incorporated into the intra-scale feature interaction mechanism to optimize the feature interaction process and improve the model’s capability of capturing global contextual key information. Finally， based on the original feature fusion method， a small object enhancement pyramid network is proposed to enhance the model’s ability to extract detailed features of small targets. Experimental results indicate that， compared to the RT-DETR model， the MCS-DETR algorithm reduces the parameter size by 20.7%， while increasing accuracy， recall， mAP₅₀， and mAP_50∶95 of 2.4%， 3.1%， 2.8%， and 2.0%， respectively on the Visdrone2019-DET-Test dataset. This method effectively migrates missed and erroneous detections in complex scenes for UAV aerial image object detection.

Key words: object detection, UAV aerial imagery, RT-DETR, multi-scale edge information enhancement, convolutional additive token mixer, small object enhancement pyramid

CLC Number:

V279

Shuai ZHONG, Liping WANG. MCS-RETR: Improved RT-DETR object detection method for UAV aerial images[J]. Acta Aeronautica et Astronautica Sinica, 2025, 46(22): 331987.

Figures/Tables 15

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Table 1

Table 2

Table 3

Table 4

Table 5

Fig.7

Fig.8

Fig.9

Fig.10

References 34

[1]	王传云，苏阳，王琳霖，等. 面向反制无人机集群的多目标连续鲁棒跟踪算法［J］. 航空学报， 2024， 45（7）： 329017.
	WANG C Y， SU Y， WANG L L， et al. Multi-object continuous robust tracking algorithm for anti-UAV swarm［J］. Acta Aeronautica et Astronautica Sinica， 2024， 45（7）： 329017 （in Chinese）.
[2]	孟凡腾，秦勇，崔京，等. 铁路外部环境无人机图像未知风险检测方法［J］. 航空学报， 2025， 46（11）： 531262.
	MENG F T， QIN Y， CUI J， et al. Unknown risk detection in external environment of railroad using UAV images［J］. Acta Aeronautica et Astronautica Sinica， 2025， 46（11）： 531262 （in Chinese）.
[3]	吴一全，童康. 基于深度学习的无人机航拍图像小目标检测研究进展［J］. 航空学报， 2025， 46（3）： 030848.
	WU Y Q， TONG K. Research advances on deep learning-based small object detection in UAV aerial images［J］. Acta Aeronautica et Astronautica Sinica， 2025， 46（3）： 030848 （in Chinese）.
[4]	LOWE D G. Distinctive image features from scale-invariant keypoints［J］. International Journal of Computer Vision， 2004， 60（2）： 91-110.
[5]	DALAL N， TRIGGS B. Histograms of oriented gradients for human detection［C］∥2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition （CVPR’05）. Piscataway： IEEE Press， 2005： 886-893.
[6]	FELZENSZWALB P F， GIRSHICK R B， MCALLESTER D， et al. Object detection with discriminatively trained part-based models［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2010， 32（9）： 1627-1645.
[7]	REN S Q， HE K M， GIRSHICK R， et al. Faster R-CNN： Towards real-time object detection with region proposal networks［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2017， 39（6）： 1137-1149.
[8]	LIN T Y， GOYAL P， GIRSHICK R， et al. Focal loss for dense object detection［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2020， 42（2）： 318-327.
[9]	KHANAM R， HUSSAIN M. What is YOLOv5： A deep look into the internal features of the popular object detector ［DB/OL］. arXiv preprint： 2407.20892， 2024.
[10]	VARGHESE R， M S. YOLOv8： A novel object detection algorithm with enhanced performance and robustness［C］∥2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems （ADICS）. Piscataway： IEEE Press， 2024： 1-6.
[11]	WANG A， CHEN H， LIU L， et al. Yolov10： Real time end-to-end object detection［DB/OL］. arXiv preprint： 2405.14458， 2024.
[12]	KHANAM R， HUSSAIN M. Yolov11： An overview of the key architectural enhancements［DB/OL］. arXiv preprint： 2410.17725， 2024.
[13]	ZHAO Y A， LV W Y， XU S L， et al. DETRs beat YOLOs on real-time object detection［C］∥2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Piscataway： IEEE Press， 2024： 16965-16974.
[14]	CARION N， MASSA F， SYNNAEVE G， et al. End-to-end object detection with transformers［C］∥Computer Vision-ECCV 2020. Cham： Springer， 2020： 213-229.
[15]	钟帅，王丽萍. 无人机航拍图像目标检测技术研究综述［J］. 激光与光电子学进展， 2025， 62（10）： 71-89.
	ZHONG S， WANG L P. Review of Research on Object Detection in UAV Aerial Images［J］. Laser & Optoelectronics Progress， 2025， 62（10）： 71-89 （in Chinese）.
[16]	SUN F X， HE N， LI R J， et al. GD-PAN： A multiscale fusion architecture applied to object detection in UAV aerial images［J］. Multimedia Systems， 2024， 30（3）： 143.
[17]	LIU S， QI L， QIN H F， et al. Path aggregation network for instance segmentation［C］∥2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE Press， 2018： 8759-8768.
[18]	何戚天，李为相，程明，等. 面向航拍图像的轻量化目标检测算法［J］. 电光与控制， 2025， 32（3）： 56-61， 81.
	HE Q T， LI W X， CHENG M， et al. A lightweight target detection algorithm for aerial images［J］. Electronics Optics & Control， 2025， 32（3）： 56-61， 81 （in Chinese）.
[19]	CHEN J R， KAO S H， HE H， et al. Run， don’t walk： Chasing higher FLOPS for faster neural networks［C］∥ 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Piscataway： IEEE Press， 2023： 12021-12031.
[20]	WOO S， PARK J， LEE J Y， et al. CBAM： Convolutional block attention module［C］∥Computer Vision-ECCV 2018. Cham： Springer， 2018： 3-19.
[21]	李子豪，王正平，贺云涛. 基于自适应协同注意力机制的航拍密集小目标检测算法［J］. 航空学报， 2023， 44（13）： 327944.
	LI Z H， WANG Z P， HE Y T. Aerial-photography dense small target detection algorithm based on adaptive cooperative attention mechanism［J］. Acta Aeronautica et Astronautica Sinica， 2023， 44（13）： 327944 （in Chinese）.
[22]	LI Z X， HE Q H， REN L F， et al. PCAF： UAV scenarios detector via pyramid converge-and-assign fusion network［J］. Multimedia Systems， 2025， 31（1）： 25.
[23]	王朝辉，严一鸣，韩晓微，等. 基于改进SRGAN的无人机航拍图像去雾算法［J］. 激光与红外， 2024， 54（6）： 991-997.
	WANG Z H， YAN Y M， HAN X W， et al. Improved SRGAN-based algorithm for defogging UAV aerial images［J］. Laser & Infrared， 2024， 54（6）： 991-997 （in Chinese）.
[24]	LEDIG C， THEIS L， HUSZÁR F， et al. Photo-realistic single image super-resolution using a generative adversarial network［C］∥2017 IEEE Conference on Computer Vision and Pattern Recognition （CVPR）. Piscataway： IEEE Press， 2017： 105-114.
[25]	HE K M， ZHANG X Y， REN S Q， et al. Deep residual learning for image recognition［C］∥2016 IEEE Conference on Computer Vision and Pattern Recognition （CVPR）. Piscataway： IEEE Press， 2016： 770-778.
[26]	WANG C Y， MARK LIAO H Y， WU Y H， et al. CSPNet： A new backbone that can enhance learning capability of CNN［C］∥2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops （CVPRW）. Piscataway： IEEE Press， 2020： 1571-1580.
[27]	VASWANI A， SHAZEER N， PARMAR N， et al. Attention is all you need ［C］∥Proceedings of the 31st International Conference on Neural Information Processing Systems. New York： ACM， 2017： 6000-6010.
[28]	ZHANG T， LI L， ZHOU Y， et al. CAS-ViT： Convolutional additive self-attention vision transformers for efficient mobile applications［DB/OL］. arXiv preprint： 2408.03703， 2024.
[29]	MEHTA S， RASTEGARI M. Separable self-attention for mobile vision transformers ［DB/OL］. arXiv preprint： 2206.02680， 2022.
[30]	SHAKER A， MAAZ M， RASHEED H， et al. Swiftformer： Efficient additive attention for transformer-based real time mobile vision applications［DB/OL］. arXiv preprint： 2303.15446， 2023.
[31]	SUNKARA R， LUO T. No more strided convolutions or pooling： A new CNN building block for low-resolution images and small objects［DB/OL］. arXiv preprint： 2208.03641， 2022.
[32]	CUI Y N， REN W Q， KNOLL A. Omni-kernel modulation for universal image restoration［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2024， 34（12）： 12496-12509.
[33]	DU D W， ZHU P F， WEN L Y， et al. VisDrone-DET2019： The vision meets drone object detection in image challenge results［C］∥2019 IEEE/CVF International Conference on Computer Vision Workshop （ICCVW） Piscataway： IEEE Press， 2019： 213-226.
[34]	WANG S， JIANG H P， YANG J X， et al. AMFEF-DETR： An end-to-end adaptive multi-scale feature extraction and fusion object detection network based on UAV aerial images［J］. Drones， 2024， 8（10）： 523.

参数	数值
epoch	200
Batch Size	4
Image Size	640
Works	4
Optimizer	Adam

模型	P/%	R/%	mAP_0.5/%	mAP_0.5∶0.95/%	Params/M	FLOPs/G
YOLOv5m	46.8	35.2	33.6	19.5	25.0	64.0
YOLOv8s	44.8	33.6	31.6	18.0	11.1	28.5
YOLOv8m	48.0	35.7	34.4	19.9	25.8	78.7
YOLOv10n	39.9	29.8	27.2	14.9	2.2	6.5
YOLOv10s	44.7	34.4	32.4	18.2	7.2	21.4
YOLOv11n	38.8	30.2	27.2	15.3	2.5	6.3
YOLOv11s	44.7	34.5	32.3	18.7	9.4	21.3
YOLOv11m	48.1	37.8	36.0	21.4	20.0	67.7
RTMDet	48.1	34.7	35.3	21.1	52.3	80.0
Efficient DETR	49.5	36.1	36.7	22.0	32.0	159
RT-DETR-R18	54.6	38.2	37.1	21.2	19.8	57.0
RT-DETR-R34	57.2	39.6	38.6	22.5	31.1	88.8
MCS-DETR	57.0	41.3	39.9	23.2	15.7	64.3

类别	P/%	R/%	mAP₅₀/%	mAP_50∶95/%
pedestrian	56.7	36.0	37.6	15.20
people	57.0	24.9	27.3	9.91
bicycle	38.7	13.6	11.7	5.18
car	76.8	75.3	76.8	49.00
van	53.1	41.5	37.5	26.30
truck	53.3	44.6	42.7	26.60
tricycle	33.3	28.5	21.5	12.10
awning-tricycle	46.3	21.3	20.2	12.40
bus	77.9	52.2	56.7	39.50
motor	53.0	43.7	39.2	16.00
all class	54.6	38.2	37.1	21.20

类别	P/%	R/%	mAP₅₀/%	mAP_50∶95/%
pedestrian	58.8	39.8	41.4	17.2
people	56.7	27.5	29.2	10.8
bicycle	42.8	16.7	14.2	5.94
car	78.3	77.8	78.9	51.3
van	55.2	42.9	38.4	27.4
truck	58.6	46.9	45.5	29.3
tricycle	36.5	36.0	27.3	15.4
awning-tricycle	49.5	23.9	22.1	13.8
bus	77.7	54.2	58.2	41.9
motor	56.2	47.2	43.9	18.9
all class	57.0	41.3	39.9	23.2

RT-DETR	MSEIE	CATM-AIFI	SOEP	Params/M	FLOPs/G	P/%	R/%	mAP_0.5/%	mAP@_0.5∶0.95/%	FPS/（帧·s^-1）
√				19 884 600	57.0	54.6	38.2	37.1	21.2	96.5
√	√			14 468 248	48.4	55.8	39.6	38.5	22.5	64.3
√		√		19 960 888	57.1	55.1	39.1	37.9	22.1	96.9
√			√	20 500 536	65.2	55.3	39.4	38.1	22.0	99.2
√	√		√	15 685 528	64.2	56.2	40.9	39.2	22.6	61.3
√	√	√	√	15 761 816	64.3	57.0	41.3	39.9	23.2	61.7

MCS-RETR: Improved RT-DETR object detection method for UAV aerial images

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 15

References 34

Related Articles 15

Recommended Articles

Metrics

Comments

[1]	Yiquan WU, Kang TONG. Research advances on deep learning-based small object detection in UAV aerial images [J]. Acta Aeronautica et Astronautica Sinica, 2025, 46(3): 30848-030848.
[2]	Yi ZHENG, Xianghong CHENG, Xingbang TANG, Yi CAO. Oriented detection algorithm for insulator and their defects from aerial images based on improved ReDet [J]. Acta Aeronautica et Astronautica Sinica, 2025, 46(18): 331825-331825.
[3]	Fanteng MENG, Yong QIN, Jing CUI, Yunpeng WU, Zicheng ZHANG, Shaowei WEI. Unknown risk detection in external environment of railroad using UAV images [J]. Acta Aeronautica et Astronautica Sinica, 2025, 46(11): 531262-531262.
[4]	Shusheng CHEN, Muliang JIA, Jiahao LIN, Shiyi JIN, Zhenghong GAO, Yueqing WANG, Zhiqiang MA, Zheng LI, Chenlong DUAN, Jiawei LI. Empowering aircraft technology applications with generative models: Research progress and prospects [J]. Acta Aeronautica et Astronautica Sinica, 2025, 46(10): 631194-631194.
[5]	Xudong LUO, Yiquan WU, Jinlin CHEN. Research progress on deep learning methods for object detection and semantic segmentation in UAV aerial images [J]. Acta Aeronautica et Astronautica Sinica, 2024, 45(6): 28822-028822.
[6]	Junyu LI, Qiankun LIU, Ying FU. Infrared small object detection based on attention mechanism [J]. Acta Aeronautica et Astronautica Sinica, 2024, 45(14): 628959-628959.
[7]	Jiqiang GAN, Xiaoping WANG. Surface defect detection of fiber placement based on virtual sample generation [J]. Acta Aeronautica et Astronautica Sinica, 2024, 45(1): 428624-428624.
[8]	Zhiqiang FENG, Zhijun XIE, Zhengwei BAO, Kewei CHEN. Real⁃time dense small object detection algorithm for UAV based on improved YOLOv5 [J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2023, 44(7): 327106-327106.
[9]	Guotao MAO, Tianmin DENG, Nanjing YU. Object detection in UAV images based on multi-scale split attention [J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2023, 44(5): 326738-326738.
[10]	Yubin YUAN, Yiquan WU, Langyue ZHAO, Jinlin CHEN, Qichang ZHAO. Research progress of UAV aerial video multi⁃object detection and tracking based on deep learning [J]. Acta Aeronautica et Astronautica Sinica, 2023, 44(18): 28334-028334.
[11]	Yucun SONG, Quanbo GE, Junlong ZHU, Zhenyu LU. Improved YOLOX object detection algorithm based on gradient difference adaptive learning rate optimization [J]. Acta Aeronautica et Astronautica Sinica, 2023, 44(14): 327951-327951.
[12]	Zihao LI, Zhengping WANG, Yuntao HE. Aerial-photography dense small target detection algorithm based on adaptive cooperative attention mechanism [J]. Acta Aeronautica et Astronautica Sinica, 2023, 44(13): 327944-327944.
[13]	LIU Fang, HAN Xiao. Adaptive aerial object detection based on multi-scale deep learning [J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2022, 43(5): 325270-325270.
[14]	WANG Hui, JIA Zikai, JIN Ren, LIN Defu, FAN Junfang, XU Chao. Cooperative object detection in UAV-based vision-guided docking [J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2022, 43(1): 324854-324854.
[15]	LI Hongguang, YU Ruonan, DING Wenrui. Research development of small object traching based on deep learning [J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2021, 42(7): 24691-024691.