基于融合注意力机制和图像识别的路桥裂缝检测技术

Road and Bridge Crack Detection Technology based on Fusion Attention Mechanism and Image Recognition

针对当前路桥检测领域面临的训练效率低下与检测性能不足的挑战,设计了基于Inception-Resnet-v2的路桥裂缝检测模型。该模型结合其强大的特征学习能力与多尺度特征融合,显著提升了对复杂路桥环境的裂缝检测精度。同时,GKA聚类算法的应用有效减少了非必要区域的计算,提高了检测效率。结果表明:与AlexNet相比,所提模型不仅在帧率(FPS)上实现了8.67%的提升,确保了实时处理的潜力,同时在准确率、精度、召回率及F1分数上分别取得了3.19%、3.75%、1.34%和2.66%的显著提升。该模型为提升路桥检测技术的智能化水平提供了有力支持,并为该领域未来的研究与发展提供了参考与借鉴。

In response to the challenges of low training efficiency and insufficient detection performance in the current field of road and bridge detection,this paper designs a road and bridge crack detection model based on Inception Resnet-v2.This model combines its powerful feature learning ability with multi-scale feature fusion,significantly improving the accuracy of crack detection in complex road and bridge environments.Meanwhile,the application of GKA clustering algorithm effectively reduces the computation of unnecessary regions and improves detection efficiency.The experimental results show that compared with AlexNet,the proposed model not only achieves an 8.67%improvement in frame rate(FPS),ensuring the potential for real-time processing,but also achieves significant improvements in accuracy,precision,recall,and F1 score of 3.19%,3.75%,1.34%,and 2.66%,respectively.This model provides strong support for improving the intelligence level of road and bridge detection technology,and provides valuable reference and inspiration for future research and development in this field.