To address issues such as unclear crack boundaries and inaccurate contour segmentation in crack detection and segmentation, an improved YOLO11n-seg-based algorithm for crack boundary detection and segmentation was proposed. An enhanced edge attention convolution(EEA) module was designed, and multi-scale atrous convolution was employed to strengthen edge-feature processing. A channel-spatial parallel attention(CSPA) module was introduced, and channel and spatial information were weighted and fused so that attention was concentrated on target regions and learning capability was enhanced. A lightweight dynamic upsampling module(DySample) was incorporated into the neck network, and feature details and semantic information were better preserved. Experiments were conducted on a self-made propellant tensile crack dataset. Compared with the original YOLO11n-seg model, improvements of 3.6% in precision, 1.6% in recall, 1.8% in mAP@0.5, and 4.6% in mAP@0.5:0.95 are achieved. In cross-model comparisons, an average improvement of 15% in mAP@0.5:0.95 over other baseline models is obtained, and the effectiveness and accuracy of the proposed method are validated. The proposed algorithm is of great significance for propellant damage evolution analysis, structural integrity evaluation, and engine safety assessment.
尽管上述研究模型在相关领域表现非常出色,但是当应用在推进剂拉伸裂纹分割时却表现出精度低、训练时间长、无法有效区分边界等问题,其综合性能无法匹配YOLO(you only look once)系列中最新的YOLO11-seg[5]。在YOLO11改进模型方面,已有部分研究取得了一定的进展,但是面向推进剂拉伸裂纹的检测与分割暂无较好的模型,王国国等[22]提出基于YOLO11的绝缘子缺陷检测,引入了GhostConv模块与CA(coordinate attention for efficient mobile network design)注意力,然而卷积层生成的特征图无法捕捉裂纹边缘的细节特征。靳子越等[23]提出了基于YOLO11-seg的轻量级岩石裂隙检测及分割算法,通过结合不同感受野的卷积层实现了对特征的多尺度筛选与增强,然而引入的Focaler-IoU(intersection over union)使得网络在达到设定阈值后不会继续优化边界精度,影响mAP@0.5:0.95。
由于模块设计时采用了并行特征提取的结构,一定程度上增加了训练的参数量,针对这个问题,引入了通道缩减参数(channel_reduce),通常设置为4,表示可以将inputs输入特征的通道数缩减为原来的1/4,通过该设计可以减少浮点运算次数(floating point operations,FLOPs),在不太影响训练结果的情况下显著提高训练速度,尤其是在深层网络结构中。计算方法为
MahjubA, MazlanN M, AbdullahM Z, et al. Design optimization of solid rocket propulsion: a survey of recent advancements[J]. Journal of Spacecraft and Rockets, 2020, 57(1): 3-11.
[2]
TangC, QiangH, GengT, et al. Advances in structural reliability analysis of solid propellant grain: a comprehensive review[J]. Polymers, 2025, 17(15): 2039.
[3]
SobelI, FeldmanG. A 3×3 isotropic gradient operator for image processing[R]. Stanford: Stanford Artificial Intelligence Project, 1968: 271-272.
[4]
CannyJ. A computational approach to edge detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, 8(6): 679-698.
[5]
MathewA, AmudhaP, SivakumariS. Deep learning techniques: an overview[C]//International Conference on Advanced Machine Learning Technologies and Applications. Singapore: Springer Singapore, 2020: 599-608.
[6]
LuoQ, FangX, LiuL, et al. Automated visual defect detection for flat steel surface: a survey[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69(3): 626-644.
[7]
LingQ, IsaN A M. Printed circuit board defect detection methods based on image processing, machine learning and deep learning: a survey[J]. IEEE Access, 2023, 11: 15921-15944.
[8]
GuoJ, LiuP, XiaoB, et al. Surface defect detection of civil structures using images: review from data perspective[J]. Automation in Construction, 2024, 158: 105186.
[9]
WangX, GaoH, JiaZ, et al. BL-YOLOv8: an improved road defect detection model based on YOLOv8[J]. Sensors, 2023, 23(20): 8361.
[10]
TaoX, ZhangD, MaW, et al. Automatic metallic surface defect detection and recognition with convolutional neural networks[J]. Applied Sciences, 2018, 8(9): 1575.
[11]
ZhangG, PanY, ZhangL. Semi-supervised learning with GAN for automatic defect detection from images[J]. Automation in Construction, 2021, 128: 103764.
[12]
QiuJ L, ShenY S, LinJ C, et al. An enhanced method for surface defect detection in workpieces based on improved MobileNetV2-SSD[J]. Expert Systems, 2025, 42(1): e13567.
ZhaoShunxiang, ZhouAnliang, WangJia, et al. Defect detection in constant-diameter growth process of monocrystalline silicon based on an improved YOLOv8n model[J]. Science Technology and Engineering, 2026, 26(4): 1565-1574.
[19]
YuZ, WuY, WeiB, et al. A lightweight and efficient model for surface tiny defect detection[J]. Applied Intelligence, 2023, 53(6): 6344-6353.
[20]
ZhengJ, SunX, ZhouH, et al. Printed circuit boards defect detection method based on improved fully convolutional networks[J]. IEEE Access, 2022, 10: 109908-109918.
[21]
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.
ZhangXuehui, LiXiaohang, TianXuezhao, et al. Segmentation method based on light measurement multi-scale dynamic fusion module[J]. Science Technology and Engineering, 2025, 25(18): 7719-7728.
[24]
NguyenH T, NguyenL T, AfanasievA D, et al. Classification of road pavement defects based on convolution neural network in Keras[J]. Automatic Control and Computer Sciences, 2022, 56(1): 17-25.
WangGuoguo, GongXiaoyu, YuanFating. Lightweight algorithm based on YOLO11s insulator defect detection[J/OL]. Journal of Chongqing Technology and Business University(Natural Science Edition), (2024-12-26)[2025-02-12].
JinZiyue, LiHaitao, YinHaichen, et al. An improved lightweight rock fracture detection and segmentation algorithm based on YOLO11n-seg[J]. Journal of Peking University(Natural Science Edition), 2026, 62(2): 237-252.
[31]
JiangP, ErguD, LiuF, et al. A review of YOLO algorithm developments[J]. Procedia Computer Science, 2022, 199: 1066-1073.
DangJianwu, ZhangTianyin, TianBin. Multi-scale fusion edge detection model with spatial co-location rule based on dense extreme inception network[J]. Journal of Hunan University(Natural Science Edition), 2024, 51(8): 13-22.
[34]
HuangH, ChenZ, ZouY, et al. Channel prior convolutional attention for medical image segmentation[J]. Computers in Biology and Medicine, 2024, 178: 108784.
[35]
LiuW, LuH, FuH, et al. Learning to upsample by learning to sample[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. New York: IEEE, 2023: 6027-6037.
MupparajuS, ThotakuraS R, ChoudaryV R R. A review on YOLOv8 and its advancements[C]//International Conference on Data Intelligence and Cognitive Informatics. Singapore: Springer, 2024: 529-545.
[38]
WangC Y, YehI H, Mark LiaoH Y. Yolov9: learning what you want to learn using programmable gradient information[C]//European Conference on Computer Vision. Cham: Springer Nature Switzerland, 2024: 1-21.
[39]
WangA, ChenH, LiuL, et al. YOLOv10: real-time end-to-end object detection[C]//Advances in Neural Information Processing Systems. Vancouver: NeurIPS, 2024: 107984-108011.
[40]
KhanamR, HussainM. YOLOv11: an overview of the key architectural enhancements[J]. arXiv Preprint. arXiv: 2410. 17725, 2024.