For the issues of overlapping obstacle targets on driving roads, inconsistent scale characteristics, and high misdetection rates due to easy confusion with the background leading to low detection accuracy, an obstacle detection algorithm ADMH-YOLOv8 was proposed. Based on the YOLOv8 framework, firstly, a variable kernel convolution was used to improve the feature extraction network’s standard convolutional layers to enhance its feature extraction capability. Secondly, combining dilated convolution to improve the dynamic sampling operator (DySample) instead of fixed upsampling enhances the effectiveness of feature fusion. Again, an improved hybrid attention module that integrates multi-scale features was introduced to reduce information loss during upsampling and enhance the perception ability of important regions. Finally, a parameter sharing detection head (PSDH) was designed to reduce the number of parameters. The experimental results show that the ADMH-YOLO algorithm has excellent comprehensive performance, with an average accuracy of 93.8%, which is 2.0% higher than the baseline algorithm. It has high obstacle recognition accuracy and is suitable for obstacle detection tasks on roads.
LI Yu-juan, GE Dong-yuan, YAO Xi-fan, ZHOU Hao-wei.
Fusion of Hybrid Attention and Detection Head for Road Obstacle Detection[J].
Science Technology and Engineering, 2026, 26(1): 253-261 DOI:10.12404/j.issn.1671-1815.2500190
采用平均精度均值(mean average pre-sion, mAP)、参数量(paremeters)和浮点运算量(GFLOPs)作为评估指标。mAP代表了对4个类别平均识别精度,值越大,表示模型的精度越好,mAP50是指当交并比IoU阈值等于0.5时,每个类别AP的平均值,mAP50-95是指当IoU阈值为0.5~0.95时,每个类别AP的平均值;较小的参数量意味着模型越小,浮点运算量越少则表示模型的复杂度越低,计算公式如下。
Global status report on road safety 2023[EB/OL]. ( 2023-12-13)[2025-01-08]. https://iris.who.int/bitstream/handle/10665/375016/9789240086517-eng.pdf?sequence=1.
[2]
李洋. 智能车辆障碍物检测技术综述[J]. 大众科技, 2019, 21(6): 65-68.
[3]
LiYang. Overview of obstacle detection technology in intelligent vehicles[J]. Volkswagen Technology, 2019, 21(6): 65-68.
ChenMingqiang, FengShujuan, ZhangYong, et al. Obstacle detection algorithm for UAV based on improved YOLOv5[J]. Science Technology and Engineering, 2024, 24(31): 13627-13634.
[6]
GirshickR, DonahueJ, DarrellT, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Columbus: IEEE, 2014: 580-587.
[7]
GirshickR. Fast R-CNN[C]// Proceedings of the IEEE International Conference on Computer Vision. Santiago: IEEE, 2015: 1440-1448.
[8]
RenS Q, HeK M, GirshickR, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6): 1137-1149.
[9]
KortmannF, TalitsK, FassmeyerP, et al. Detecting various road damage type sing lobal countries utilizing Faster R-CNN[C]// IEEE International Conference on Big Data (Big Data). New York: IEEE, 2020: 5563-5571.
[10]
ChaY J, ChoiW, SuhG, et al. Autonomous structural visual inspection using region-based deep learning for detecting multiple damage types[J]. Computer-aided Civil and Infrastructure Engineering, 2018, 33(9): 731-747.
[11]
LiuW, AnguelovD, ErhanD, et al. SSD: single shot multibox detector[C]// Proceedings of the Computer Vision ECCV 2016: 14th European Conference. Amsterdam: IEEE, 2016: 21-37.
[12]
LinT Y, GoyalP, GirshickR, et al. Focal loss for dense object detection[EB/OL]. ( 2017-08-07) [2025-01-08]. https://arxiv.org/abs/1708.02002.
[13]
RedmonJ, DivvalaS, GirshickR, et al. You only look once: unified, real-time object detection[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 779-788.
[14]
RedmonJ, FarhadiA. YOLO9000: better, faster, stronger[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 7263-7271.
[15]
RedmonJ, FarhadiA. YOLOv3: an incremental improvement[J]. ArXiv: 1804. 02767, 2018.
[16]
BochkovskiyA, WangC Y, LiaoH Y M. YOLOv4: optimal speed and accuracy of object detection[J]. ArXiv: 2004. 10934, 2020.
LiuChunlei, LiZhihua, WangChao, et al. A pedestrian detection method integrating RepVGG and YOLOv5[J]. Science Technology and Engineering, 2023, 23(7): 2945-2951.
WangHong, HanChen, YuanBoyang, et al. Improved YOLOv5-based pedestrian crowding detection algorithm[J]. Science Technology and Engineering, 2023, 23(27): 11730-11738.
[21]
LiC Y, LiL L, JiangH L, et al. YOLOv6: a single-stage object detection framework for industrial applications[J]. ArXiv: 2209. 02976, 2022.
[22]
WangC Y, BochkovskiyA, LiaoH Y M. YOLOv7: trainable bag-of-freebies sets new state of the art for real time object detectors[C]// Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver: IEEE, 2023: 7464-7475.
[23]
WangC Y, YehI H, LiaoH Y M. YOLOv9: learning what you want to learn using programmable gradient information[J]. ArXiv: 2402. 13616, 2024.
[24]
WangA, ChenH, LiuL H, et al. YOLOv10: real-time end to end object detection[J]. ArXiv: 2405. 14458, 2024.
WangHaodong, WangLiyong, SuQinghua, et al. Road identification algorithm for temporary roads based on YOLOv4 network model[J]. Journal of Guangxi University (Natural Science Edition), 2022, 47(3): 712-722.
[27]
许洁. 基于深度学习的路面质量评估的研究和实现[D]. 成都: 电子科技大学, 2022.
[28]
XuJie. Research and implementation on road quality assessment based on deep learning[D]. Chengdu: University of Electronic Science and Technology of China, 2022.
NiChangshuang, LiLin, LuoWenting, et al. Disease detection of asphalt pavement based on improved YOLOv7[J]. Computer Engineering and Applications, 2023, 59(13): 305-316.
XuanYiguo, YuChengbo, JiangQichao, et al. Improved YOLOv7 road crack and pit detection algorithm[J]. Science Technology and Engineering, 2024, 24(17): 7205-7213.
[33]
LiT, LiG. Road defect identification and location method based on an improved ML-YOLO algorithm[J]. Sensors, 2024, 24(21): 6783.
[34]
ZhangX, SongY Z, SongT T, et al. AKConv: convolutional kernel with arbitrary sampled shapes and arbitrary number of parameters[J]. ArXiv: 2311. 11587, 2023.
[35]
LiuW Z, LuH, FuH T, et al. Learning to upsample by learning to sample[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision. Paris: IEEE, 2023: 6027-6037.
[36]
WooS, ParkJ, LeeJ Y, et al. CBAM:convolutional block attention module[C]// Proceedings of the 15th European Conference on Computer Vision (ECCV). Munich: Springer, 2018: 3-19.
[37]
TianZ, ShenC H, ChenH, et al. FCOS: a simple and strong anchor-free object detector[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44: 1922-1933.
[38]
ChenZ X, HeZ W, LuZ M. DEA-Net: single image dehazing based on detail-enhanced convolution and content-guided attention[J]. IEEE Transactions on Image Processing, 2024, 33(2): 1002-1015.