基于改进YOLOv8的轻量化无人机图像目标检测算法

doi:10.19678/j.issn.1000-3428.0070085

摘要/Abstract

摘要： 针对无人机图像中小目标实例多、目标间存在遮挡所导致的漏检、误检等现象，提出了一种基于改进YOLOv8的轻量化无人机图像小目标检测算法。在颈部引入三特征编码器、尺度序列特征融合模块，增强了网络对不同尺度特征的提取能力；同时，设计了小目标检测层，并与改进的颈部特征提取网络进行融合，在头部引入一个额外的检测头，减小小目标特征的损失，增强网络对小目标的识别能力；此外，针对CIoU的缺陷，结合Wise-IoU、inner-IoU和MPDIoU，提出了一种回归损失算法Wise-inner-MPDIoU；最后，为了实现算法在移动端和嵌入式场景下的轻量化应用需求，进行了基于幅度的层自适应稀疏化剪枝，在保证模型精度的同时，进一步压缩了模型的大小。实验结果表明，改进后的模型相比于原YOLOv8s模型，在mAP50提高6.8%的同时，参数量下降76.1%，计算量下降17.1%，模型大小降低73.5%，达到了优秀的提升模型检测精度及轻量化效果，具有很强的实用意义。

Abstract: In view of the phenomenon of missed detection and false detection caused by many small target instances and occlusion between targets in drone images, this paper proposes a lightweight small target detection algorithm in UAV images based on improved YOLOv8. The triple feature encoder and scale sequence feature fusion module are introduced in the neck to enhance the network's ability to extract features of different scales. Furthermore, a small target detection layer is designed and fused with the improved neck feature extraction network, and an additional detection head is introduced in the head to reduce the loss of small target features and enhance the network's recognition ability of small targets. Additionally, in view of the defects of CIoU, a regression loss algorithm Wise-inner-MPDIoU is proposed by combining Wise-IoU, inner-IoU and MPDIoU. Finally, in order to realize the lightweight application requirements of the algorithm in mobile and embedded scenarios, amplitude-based layer adaptive sparse pruning is performed to further compress the model size while ensuring the accuracy of the model. Experimental results demonstrate that compared with the original YOLOv8s model, the improved model this paper proposed improves mAP50 by 6.8%, while the number of parameters decreases by 76.1%, the amount of computation decreases by 17.1%, and the model size decreases by 73.5%. This improved model achieves excellent accuracy improvement, lightweight effect, and practical value.

唐克, 魏飞鸣, 李东瀛, 郁文贤. 基于改进YOLOv8的轻量化无人机图像目标检测算法[J]. 计算机工程, doi: 10.19678/j.issn.1000-3428.0070085.

Tang Ke, Wei Feiming, Li Dongying, Yu Wenxian. Lightweight Target Detection Algorithm in UAV Images Based on Improved YOLOv8[J]. Computer Engineering, doi: 10.19678/j.issn.1000-3428.0070085.

参考文献

[1] Saeed Z, Yousaf M H, Ahmed R, et al. On-board small-scale object detection for unmanned aerial vehicles (UAVs)[J]. Drones, 2023, 7(5): 310.
[2] Bisio I, Haleem H, Garibotto C, et al. Performance evaluation and analysis of drone-based vehicle detection techniques from deep learning perspective[J]. IEEE Internet of Things Journal, 2021, 9(13): 10920-10935.
[3] Hoshino W, Seo J, Yamazaki Y. A study for detecting disaster victims using multi-copter drone with a thermographic camera and image object recognition by SSD[C]//2021 IEEE/ASME international conference on advanced intelligent mechatronics (AIM). IEEE, 2021: 162-167.
[4] Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 779-788.
[5] Jiang P, Ergu D, Liu F, et al. A Review of Yolo algorithm developments[J]. Procedia computer science, 2022, 199: 1066-1073.
[6] Pirasteh S, Rashidi P, Rastiveis H, et al. Developing an algorithm for buildings extraction and determining changes from airborne LiDAR, and comparing with R-CNN method from drone images[J]. Remote Sensing, 2019, 11(11): 1272.
[7] Seo D M, Woo H J, Kim M S, et al. Identification of asbestos slates in buildings based on faster region-based convolutional neural network (faster R-CNN) and drone-based aerial imagery[J]. Drones, 2022, 6(8): 194.
[8] Liu W, Qiang J, Li X, et al. UAV image small object detection based on composite backbone network[J]. Mobile Information Systems, 2022, 2022(1): 7319529.
[9] Wang G, Chen Y, An P, Hong H, Hu J, Huang T. UAV-YOLOv8: A Small-Object-Detection Model Based on Improved YOLOv8 for UAV Aerial Photography Scenarios. Sensors. 2023; 23(16):7190.
[10] Guo J, Lou H, Chen H, et al. A new detection algorithm for alien intrusion on highway[J]. Scientific reports, 2023, 13(1): 10667.
[11] Zhao Tong, Liu Jieyu, and Duan Zhiqiang, “UAV Target Detection based on RetinaNet,” 2019 Chinese Control And Decision Conference (CCDC), pp. 3342–3346, Jun. 2019.
[12] 王殿伟, et al."基于改进Double-Head RCNN的无人机航拍图像小目标检测算法."北京航空航天大学学报 1-10.doi:10.13700/j.bh.1001-5965.2022.0591. Wang D W, et al. Small target detection algorithm of UAV aerial image based on improved Double-Head RCNN. Journal of Beijing University of Aeronautics and Astronautics
[13] Wang F, Wang H, Qin Z, et al. UAV target detection algorithm based on improved YOLOv8[J]. IEEE Access, 2023.
[14] 王舒梦,徐慧英,朱信忠,等.基于改进YOLOv8n航拍轻量化小目标检测算法：PECS-YOLO[J/OL].计算机工程:1-16[2024-08-27].https://doi.org/10.19678/j.issn.1000-3428.0069353. Wang S M, Xu H Y, Zhu X Z, et al. Lightweight small object detection algorithm based on improved YOLOv8n aerial photography: PECS-YOLO[J/OL]. Computer Engineering: 1-16[2024-08-27].https://doi.org/10.19678/j.issn.1000-3428.0069353.
[15] 张佳承,韦锦,陈义时.改进YOLOv8的实时轻量化鲁棒绿篱检测算法[J/OL].计算机工程:1-16[2024-08-27].https://doi.org/10.19678/j.issn.1000-3428.0069524. Zhang J C, Wei J, Chen Y S. Improved YOLOv8 real-time lightweight robust hedge detection algorithm[J/OL]. Computer Engineering: 1-16[2024-08-27].https://doi.org/10.19678/j.issn.1000-3428.0069524.
[16] Gale T, Elsen E, Hooker S. The state of sparsity in deep neural networks[J]. arXiv preprint arXiv:1902.09574, 2019.
[17] Evci U, Gale T, Menick J, et al. Rigging the lottery: Making all tickets winners[C]//International conference on machine learning. PMLR, 2020: 2943-2952.
[18] Kang M, Ting C M, Ting F F, et al. ASF-YOLO: A novel YOLO model with attentional scale sequence fusion for cell instance segmentation[J]. Image and Vision Computing, 2024, 147: 105057.
[19] Wang C Y, Bochkovskiy A, Liao H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023: 7464-7475.
[20] Liu S, Qi L, Qin H, et al. Path aggregation network for instance segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 8759-8768.
[21] Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2117-2125.
[22] Zheng Z, Wang P, Liu W, et al. Distance-IoU loss: Faster and better learning for bounding box regression[C]//Proceedings of the AAAI conference on artificial intelligence. 2020, 34(07): 12993-13000.
[23] Tong Z, Chen Y, Xu Z, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[J]. arXiv preprint arXiv:2301.10051, 2023.
[24] Zhang H, Xu C, Zhang S. Inner-iou: more effective intersection over union loss with auxiliary bounding box[J]. arXiv preprint arXiv:2311.02877, 2023.
[25] Siliang M, Yong X. MPDIoU: A loss for efficient and accurate bounding box regression[J]. arXiv preprint arXiv:2307.07662, 2023.
[26] Lee J, Park S, Mo S, et al. Layer-adaptive sparsity for the magnitude-based pruning[J]. arXiv preprint arXiv:2010.07611, 2020.
[27] Howard A G, Zhu M, Chen B, et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications[J]. arXiv preprint arXiv:1704.04861, 2017.
[28] Du D, Zhu P, Wen L, et al. VisDrone-DET2019: The vision meets drone object detection in image challenge results[C]//Proceedings of the IEEE/CVF international conference on computer vision workshops. 2019: 0-0.
[29] 张智,易华挥,郑锦.聚焦小目标的航拍图像目标检测算法[J].电子学报,2023,51(04):944-955. Zhang Z, Yi H H, Zheng J. Target detection Algorithm in aerial image focusing on small target [J]. Acta Electronica Sinica, 2023,51(04):944-955.

选择文件类型/文献管理软件名称

选择包含的内容