Research on a U-Net Bridge Crack Identification and Feature-Calculation Methods Based on a CBAM Attention Mechanism

Crack detection on bridges is an important part of assessing whether a bridge safe for service. The methods using manual inspection and bridge-inspection vehicles have disadvantages, such as low efficiency affecting road traffic. We conducted in-depth study bridge-crack proposed crack identification algorithm Unet, called the CBAM-Unet algorithm. CBAM (Convolutional Block Attention Module) lightweight convolutional attention module that combines channel (CAM) spatial (SAM), which use mechanism spatially, respectively. takes into account characteristics cracks. When used, ability to express shallow feature information enhanced, making identified cracks more complete accurate. Experimental results show can achieve accuracy 92.66% identification. used Gaussian fuzzy, Otsu medial skeletonization algorithms realise post-processing image obtain skeleton map. A measurement based skeletonised proposed, completes maximum width length with errors 1–6% 1–8%, respectively, meeting standard. extraction we present, CBAM-Unet, effectively crack-identification task, obtained segmentation parameter calculation meet standards requirements. This method greatly improves accuracy, reduces costs efficiency.