Abstract
In the field of military and security, the vehicle target has high capture value because of its high mobility, large load and high threat. In recent years, with the development of UAV technology and the improvement performance, as well as the rapid development and wide application of deep convolutional neural network in the field of object detection, vehicle detection based on UAV has become an important task, which has application value and practical significance that cannot be ignored. Compared with the traditional ground monitoring system, the UAV has high mobility, the background of the aerial image is more complex, and the target scales vary widely. Therefore, based on deep convolutional neural network, this paper proposes a feature fusion method combining context, which enhances the feature extraction ability of small targets in complex background. Then, a multi-scale region proposal matching method based on location clustering is proposed to enhance the matching degree between target features and region proposal. Finally, this paper designs a lightweight backbone based on DenseNet [1], which effectively reduces the computational cost. Experimental results show that our model reached the accuracy of 86.7% on PASCAL VOC (only car and bus are selected) dataset, which has a 3.1% bonus compared with the baseline SSD [2]. Moreover, in order to show the effect on aerial images, an additional experiment is conducted on the UAV123 dataset, which gets a 5.7% bonus. The speed of our model has a 180 fps on the NVIDIA RTX 2080.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Huang, G., Liu, Z., Laurens, V., et al.: Densely connected convolutional networks. IEEE Computer Society (2016)
Liu, W., et al.: SSD: Single shot multibox detector. In: Leibe, B., Matas, J., Sebe, N., Welling, M. (eds.)  ECCV 2016. LNCS, vol. 9905, pp. 21–37. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-46448-0_2
Girshick, R., Donahue, J., Darrell, T., et al.: Rich feature hierarchies for accurate object detection and semantic segmentation. IEEE Computer Society (2013)
Girshick, R.: Fast R-CNN. Computer Science (2015)
Ren, S., He, K., Girshick, R., et al.: Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Trans. Pattern Anal. Mach. Intell. 39(6), 1137–1149 (2017)
Redmon, J., Divvala, S., Girshick, R., et al.: You Only Look Once: Unified, Real-Time Object Detection. IEEE (2016)
Redmon, J., Farhadi A.: YOLO9000: Better, Faster, Stronger, pp. 6517–6525. IEEE (2017)
Lin, T. Y., Goyal, P., Girshick, R., et al.: Focal loss for dense object detection. IEEE Trans. Pattern Anal. Mach. Intell. (99), 2999–3007 (2017)
Deng C., He S., Han Y., et al.: Learning dynamic spatial-temporal regularization for UAV object tracking. IEEE Signal Process. Lett. (28), 1230–1234 (2019)
Zhao B., Zhao B., Tang L., et al.: Deep spatial-temporal joint feature representation for video object detection. Sensors 18(3), 774 (2018)
Han Y., Deng C., Zhang Z., et al.: Adaptive feature representation for visual tracking. In: 2017 IEEE International Conference on Image Processing, pp. 1867–1870 (2017)
Fu, C, Y., Liu, W., Ranga, A., et al.: DSSD: Deconvolutional Single Shot Detector (2017)
Zhao, Z., Han, Y., et al.: A Reliable and Real-Time Tracking Method with Color Distribution. Sensors 17(10), 2303 (2018)
Zhao B., Han Y., Wang H., et al.: Robust shadow tracking for video SAR. IEEE Geosci Remote Sens. Lett. 18(5), 821–825 (2021)
Lin, T. Y., Dollar, P., Girshick R., et al.: Feature pyramid networks for object detection. In:Â 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017)
Tang, L., Tang, W., Han, Y., et al.: A scale-aware pyramid network for multi-scale object detection in SAR images. Remote Sens. 14 (4), 973 (2022)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 Chinese Institute of Command and Control
About this paper
Cite this paper
Wu, C., Liang, R., He, S., Wang, H. (2022). Real-Time Vehicle Detection Method Based on Aerial Image in Complex Background. In: Proceedings of 2022 10th China Conference on Command and Control. C2 2022. Lecture Notes in Electrical Engineering, vol 949. Springer, Singapore. https://doi.org/10.1007/978-981-19-6052-9_46
Download citation
DOI: https://doi.org/10.1007/978-981-19-6052-9_46
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-19-6051-2
Online ISBN: 978-981-19-6052-9
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)