Aerial Target Detection Algorithm Fused with Multi-scale Features

doi:10.16182/j.issn1004731x.joss.24-0170

Abstract

Abstract:

In order to solve the problem that UAV aerial images have a large number of small target samples but little extractable feature information, which is not conducive to improving the accuracy of aerial target detection, an improved small target detection algorithm for aerial photography based on YOLOv8s is proposed. The algorithm applies deformable convolution to the feature extraction module of the backbone network to adaptively capture the details of the target at different locations and scales. The feature information at different scales of the backbone network is extracted and enhanced by the feature collection module in the multilevel information fusion function block to obtain fine global features.The context-rich semantic information is injected into the small target detection layer using the information fusion module to realize the fusion of local and global information, and the fused features are input into the detection network to obtain the detection results. The experimental results demonstrate that the proposed algorithm improves the recognition average accuracy and recall by 14.96% and 10.85%, respectively, compared to the baseline model. The leakage rate is reduced by 7.11%. The improved target detection algorithm has the highest average accuracy for small target detection compared to mainstream detection algorithms.

Key words: aerial image, deformable convolution, small target detection, multi-scale feature fusion, object detection layer

CLC Number:

TP751

Yang Lu, Pei Junying. Aerial Target Detection Algorithm Fused with Multi-scale Features[J]. Journal of System Simulation, 2025, 37(6): 1486-1498.

Figures/Tables 15

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Table 1

Fig. 8

Table 2

Fig. 9

Table 3

Ablation experiments results of improved model

模型	精度	召回率	漏检率	mAP50/%	mAP50-95/%	$P A, S$ /%	$P A, M$ /%	$P A, L$ /%	模型参数量	权重文件/MB	帧率/(帧/s)
YOLOv8s	50.2	39.6	60.4	40.1	24.1	14.6	30.5	41.5	11129454	23.5	173.9
YOLOv8s+D2f	52.0	41.0	59.0	41.9	25.1	16.8	34.1	42.8	10409717	26.9	158.7
YOLOv8s+D2f+P2	55.0	44.1	56.9	45.8	27.9	18.5	35.8	43.5	10134783	20.7	157.5
Ours	55.7	43.9	56.1	46.1	27.9	21.8	26.3	45.1	11715768	24.0	130.3

Table 3

Fig. 10

Table 4

Fig. 11

References 30

1	高新波, 莫梦竟成, 汪海涛, 等. 小目标检测研究进展[J]. 数据采集与处理, 2021, 36(3): 391-417.
	Gao Xinbo, Mo Mengjingcheng, Wang Haitao, et al. Recent Advances in Small Object Detection[J]. Journal of Data Acquisition and Processing, 2021, 36(3): 391-417.
2	Chen Chengcheng, Zeng Weiming, Zhang Xiliang. HFPNet: Super Feature Aggregation Pyramid Network for Maritime Remote Sensing Small-object Detection[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 16: 5973-5989.
3	Zhang Rui, Xie Cong, Deng Liwei. A Fine-grained Object Detection Model for Aerial Images Based on YOLOv5 Deep Neural Network[J]. Chinese Journal of Electronics, 2023, 32(1): 51-63.
4	Redmon J, Farhadi A. YOLOV3: An Incremental Improvement[C]//IEEE Conference on Computer Vision and Pattern Recognition IEEE. Piscataway: IEEE, 2018: 1-6.
5	Liu Wentao, Zhou Bin, Wang Zhangyu, et al. FPPNet: A Fixed-perspective-perception Module for Small Object Detection Based on Background Difference[J]. IEEE Sensors Journal, 2023, 23(10): 11057-11069.
6	Du Xiaoqiang, Cheng Hongchao, Ma Zenghong, et al. DSW-YOLO: A Detection Method for Ground-planted Strawberry Fruits Under Different Occlusion Levels[J]. Computers and Electronics in Agriculture, 2023, 214: 108304.
7	赵文清, 康怿瑾, 赵振兵, 等. 改进YOLOv5s的遥感图像目标检测[J]. 智能系统学报, 2023, 18(1): 86-95.
	Zhao Wenqing, Kang Yijin, Zhao Zhenbing, et al. A Remote Sensing Image Object Detection Algorithm with Improved YOLOv5s[J]. CAAI Transactions on Intelligent Systems, 2023, 18(1): 86-95.
8	Ding Qiqi, Li Peng, Yan Xuefeng, et al. CF-YOLO: Cross Fusion YOLO for Object Detection in Adverse Weather with a High-quality Real Snow Dataset[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(10): 10749-10759.
9	王银, 王飞翔, 孙前来. 多尺度特征融合车辆检测方法[J]. 系统仿真学报, 2022, 34(6): 1219-1229.
	Wang Yin, Wang Feixiang, Sun Qianlai. Vehicle Detection Method Based on Multi Scale Feature Fusion[J]. Journal of System Simulation, 2022, 34(6): 1219-1229.
10	Gao Zhengzhong, Yu Minghang, You Zhenhuan, et al. Lightweight Infrared Target Detection Algorithm based on Adaptive Feature Enhancement[C]//2023 4th International Conference on Big Data & Artificial Intelligence & Software Engineering (ICBASE). Piscataway: IEEE, 2023: 231-237.
11	Tan Mingxing, Pang Ruoming, Le Q V. EfficientDet: Scalable and Efficient Object Detection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2020: 10778-10787.
12	Wang Chengcheng, He Wei, Nie Ying, et al. Gold-YOLO: Efficient Object Detector via Gather-and-distribute Mechanism[EB/OL]. (2023-10-23) [2023-11-5]. .
13	鞠默然, 罗海波, 王仲博, 等. 改进的YOLO V3算法及其在小目标检测中的应用[J]. 光学学报, 2019, 39(7): 245-252.
	Ju Moran, Luo Haibo, Wang Zhongbo, et al. Improved YOLO V3 Algorithm and Its Application in Small Target Detection[J]. Acta Optica Sinica, 2019, 39(7): 245-252.
14	Luo Yihao, Cao Xiang, Zhang Juntao, et al. CE-FPN: Enhancing Channel Information for Object Detection[J]. Multimedia Tools and Applications, 2022, 81(21): 30685-30704.
15	Liu Zhenyu, Yang Benyi, Duan Guifang, et al. Visual Defect Inspection of Metal Part Surface via Deformable Convolution and Concatenate Feature Pyramid Neural Networks[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69(12): 9681-9694.
16	Wang Wenhai, Dai Jifeng, Chen Zhe, et al. InternImage: Exploring Large-scale Vision Foundation Models with Deformable Convolutions[C]//2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2023: 14408-14419.
17	Dumoulin Vincent, Visin Francesco. A Guide to Convolution Arithmetic for Deep Learning[EB/OL]. (2018-01-11) [2023-10-13]. .
18	徐坚, 谢正光, 李洪均. 特征平衡的无人机航拍图像目标检测算法[J]. 计算机工程与应用, 2023, 59(6): 196-203.
	Xu Jian, Xie Zhengguang, Li Hongjun. Feature-balanced UAV Aerial Image Target Detection Algorithm[J]. Computer Engineering and Applications, 2023, 59(6): 196-203.
19	Dai Yimian, Wu Yiquan, Zhou Fei, et al. Asymmetric Contextual Modulation for Infrared Small Target Detection[C]//2021 IEEE Winter Conference on Applications of Computer Vision (WACV). Piscataway: IEEE, 2021: 949-958.
20	Li Hulin, Li Jun, Wei Hanbing, et al. Slim-neck by GSConv: A Lightweight-design for Real-time Detector Architectures[J]. Journal of Real-Time Image Processing, 2024, 21(3): 62.
21	Li Chuyi, Li Lulu, Jiang Hongliang, et al. YOLOv6: A Single-stage Object Detection Framework for Industrial Applications[EB/OL]. (2022-09-07) [2023-10-14]. .
22	Ding Xiaohan, Zhang Xiangyu, Ma Ningning, et al. RepVGG: Making VGG-style ConvNets Great Again[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2021: 13728-13737.
23	Su Zihang, Yu Jiong, Tan Haotian, et al. MSA-YOLO: A Remote Sensing Object Detection Model Based on Multi-scale Strip Attention[J]. Sensors, 2023, 23(15): 6811.
24	Zhang Xiangyu, Zhou Xinyu, Lin Mengxiao, et al. ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 6848-6856.
25	Wang Kuilin, Liu Zhenze. BA-YOLO for Object Detection in Satellite Remote Sensing Images[J]. Applied Sciences, 2023, 13(24): 13122.
26	Du Dawei, Zhu Pengfei, Wen Longyin, 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, 2019: 213-226.
27	Selvaraju R R, Das A, Vedantam R, et al. Grad-CAM: Why Did You Say That?[EB/OL]. (2017-01-25) [2023-10-21]. .
28	Zhu Xingkui, Shuchang Lü, Wang Xu, et al. TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarios[C]//2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). Piscataway: IEEE, 2021: 2778-2788.
29	Wang Tianyu, Ma Zhongjing, Yang Tao, et al. PETNet: A YOLO-based Prior Enhanced Transformer Network for Aerial Image Detection[J]. Neurocomputing, 2023, 547: 126384.
30	Faraji H, Chen Bingcai. Drone-YOLO: Improved YOLO for Small Object Detection in UAV[C]//2023 8th International Conference on Image, Vision and Computing (ICIVC). Piscataway: IEEE, 2023: 93-100.

参数	实验设置
输入图像尺寸/像素	640×640
优化器	随机梯度下降(SGD)
初始学习率	0.001
动量	0.9
权重衰减	0.004 7
批处理大小	4

模型	平均精度										mAP50
模型	行人	人	自行车	汽车	卡车	面包车	三轮车	遮阳敞篷	公共汽车	摩托	mAP50
YOLOv5s	39.0	31.9	11.2	72.4	34.4	32.0	20.1	12.3	40.7	38.6	33.3
YOLOv7	20.9	14.8	7.3	51.0	29.7	19.5	14.0	8.8	30.5	21.2	41.4
YOLOv8s	41.8	32.6	13.0	79.7	46.3	39.3	29.9	16.9	57.0	44.1	40.1
TPH-YOLOv5^[28]	29.0	16.7	15.6	68.9	49.7	45.1	27.3	24.7	61.8	30.9	37.3
PETNet^[29]	34.8	23.1	20.4	65.0	40.7	37.2	27.6	15.3	54.4	33.3	35.3
Drone-YOLO^[30]	26.9	14.2	16.0	65.7	42.8	47.7	21.4	19.3	57.1	25.9	39.1
本文	52.6	44.1	19.0	85.1	50.8	41.5	32.9	19.3	61.4	54.4	46.1

[1]	Guo Yecai, Sun Jingdong, Saha Amitave. Improved Target Detection Algorithm for Aerial Images Based on YOLOv5 [J]. Journal of System Simulation, 2025, 37(2): 551-562.
[2]	Zhang Xiliu, Zhang Xiaoling, He Minjun. Research on Vehicle Detection Method Based on Improved YOLOX-s [J]. Journal of System Simulation, 2024, 36(2): 487-496.
[3]	Cheng Haiyan, Han Pu, Wang Di, Zhai Yongjie. Location Method of Insulators in Power Grid Patrol Aerial Images [J]. Journal of System Simulation, 2017, 29(6): 1327-1336.