改进YOLOv7的航拍图像目标检测

doi:10.3778/j.issn.1002-8331.2312-0100

摘要/Abstract

摘要： 针对无人机航拍图像中尺度变化、小目标且密集遮挡情况导致检测精度降低的问题，提出一种改进YOLOv7的航拍图像目标检测算法。设计一种联合动态卷积的加权采样模块，从多个维度捕捉特征，提高模型特征提取能力；增加浅层特征检测头，保留更多的细节信息，增强对小目标特征的利用能力；在特征融合部分构建一种具有残差结构的多尺度特征聚合模块（C2-Res2Block），使模型融合丰富的多尺度信息。使用MPDIoU度量替换传统IOU计算边界回归损失，提高模型对密集遮挡目标的定位能力。通过在无人机航拍数据集VisDrone2019上进行实验表明，改进后的算法较原模型mAP@0.5提高了4.3个百分点，mAP@0.5：0.95提高了2.4个百分点，参数量减少了6.81×106，与目前主流的目标检测算法相比也取得更高的检测精度，有效提高了对无人机航拍图像的检测精度，并明显改善对航拍目标的误检和漏检情况。

关键词: 目标检测, YOLOv7, 动态卷积, 检测头, C2-Res2Block, MPDIoU

Abstract: An improved YOLOv7 aerial image object detection algorithm is proposed to solve the problems of low detection accuracy caused by mesoscale changes, small targets and dense occlusion in UAV aerial images. Firstly, a weighted sampling module with joint dynamic convolution is designed to capture features from multiple dimensions and improve the feature extraction ability of the model. Secondly, add a shallow feature detection head to retain more detailed information and enhance the ability to utilize small target features. Then, a multi-scale feature aggregation module (C2-Res2Block) with residual structure is constructed in the feature fusion part to make the model fuse rich multi-scale information. Finally, the MPDIoU measure is used to replace the traditional IOU to calculate the boundary regression loss and improve the localization ability of the model to the densely occluding target. Experiments on UAV aerial photography data set VisDrone2019 show that the improved algorithm is 4.3 percentage points higher than the original model on mAP@0.5, 2.4 percentage points on mAP@0.5：0.95, the number of parameters is reduced by 6.81×106, and the detection accuracy is higher than the current mainstream object detection algorithms. It effectively improves the detection accuracy of UAV aerial images, and obviously improves the false detection and missing detection of aerial objects.

Key words: object detection, YOLOv7, dynamic convolution, detection head, C2-Res2Block, MPDIoU

崔丽群, 曹华维. 改进YOLOv7的航拍图像目标检测[J]. 计算机工程与应用, 2024, 60(20): 189-197.

CUI Liqun, CAO Huawei. Object Detection Based on Improved YOLOv7 for UAV Aerial Image[J]. Computer Engineering and Applications, 2024, 60(20): 189-197.

参考文献

[1] 张华卫, 张文飞, 蒋占军, 等. 引入上下文信息和Attention Gate的GUS-YOLO遥感目标检测算法[J]. 计算机科学与探索, 2024, 18(2): 453-464.
ZHANG H W, ZHANG W F, JIANG Z J, et al. GUS-YOLO remote sensing target detection algorithm introducing context information and attention gate[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(2): 453-464.
[2] 王春梅, 刘欢. YOLOv8-VSC: 一种轻量级的带钢表面缺陷检测算法[J]. 计算机科学与探索, 2024, 18(1): 151-160.
WANG C M, LIU H. YOLOv8-VSC: lightweight algorithm for strip surface defect detection[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(1): 151-160.
[3] 周颖, 颜毓泽, 陈海永, 等. 基于改进YOLOv8的光伏电池缺陷检测[J]. 激光与光电子学进展, 2024, 61(8): 235-245.
ZHOU Y, YAN Y Z, CHEN H Y, et al. Defect detection of photovoltaic cells based on improved YOLOv8[J]. Laser & Optoelectronics Progress, 2024, 61(8): 235-245.
[4] GIRSHICK R. Fast R-CNN[C]//Proceedings of the International Conference on Computer Vision, 2015: 1440-1448.
[5] REN S Q, HE K, 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.
[6] 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.
[7] REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 7263-7271.
[8] REDMON J, FARHADI A. YOLOv3: an incremental improvement[C]//Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition, 2018: 1-6.
[9] BOCHKOVSKIY A, WANG C Y, LIAO H. YOLOv4: optimal speed and accuracy of object detection[J]. arXiv:2004.10934, 2020.
[10] LI C, LI L, JIANG H, et al. YOLOv6: a single stage object detection framework for industrial applications[J]. arXiv:2209.02976, 2022.
[11] 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[J]. arXiv:2207.02696, 2022.
[12] LIU Z, GA G, SUN L, et al. HRDNet: high-resolution detection network for small objects[C]//Proceedings of the 2021 IEEE International Conference on Multimedia and Expo, 2021: 1-6.
[13] 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 J, XIE Z G, LI H J. Feature-balanced UAV aerial image target detection algorithm[J]. Computer Engineering and Applications, 2023, 59(6): 196-203.
[14] ZHU X, LYU S, WANG X, et al. TPH-YOLOv5: improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 2778-2788.
[15] 张灵灵, 王鹏, 李晓艳, 等. 基于优化SSD的低空无人机检测方法[J]. 计算机工程与应用, 2022, 58(16): 204-212.
ZHANG L L, WANG P, LI X Y, et al. Low-altitude UAV detection method based on optimized SSD[J]. Computer Engineering and Applications, 2022, 58(16): 204-212.
[16] HUANG H, LI L, MA H. An improved cascade R-CNN-based target detection algorithm for UAV aerial images[C]//Proceedings of the 2022 7th International Conference on Image, Vision and Computing, 2022: 232-237.
[17] 白宗宝, 张俊举, 高原, 等. 基于注意力机制的航拍图像目标检测算法[J]. 激光与光电子学进展, 2023, 60(12): 312-322.
BAI Z B, ZHANG J J, GAO Y, et al. Attention mechanism-based object detection algorithm in aerial images[J]. Laser & Optoelectronics Progress, 2023, 60(12): 312-322.
[18] 王晓红, 胡豫. 复杂背景下的无人机图像小目标检测[J]. 计算机工程与应用, 2023, 59(15): 107-114.
WANG X H, HU Y. UAV image small object detection on complex background[J]. Computer Engineering and Applications, 2023, 59(15): 107-114.
[19] CHAO L, AOJUN Z, ANBANG Y. Omni-dimensional dynamic convolution[J]. arXiv:2209.07947, 2022.
[20] GAO S H, CHENG M M, ZHAO K, et al. Res2Net: a new multi-scale backbone architecture[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 43(2): 652-662.
[21] SILIANG M, YONG X. MPDIoU: a loss for efficient and accurate bounding box regression[J]. arXiv:2307.07662, 2023.
[22] 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: 213-226.
[23] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision, 2016: 21-37.