A Computer Vision-Aided Lightweight Attention-Augmented Approach for Nondestructive Steel Wire Rope Defect Detection

Steel wire rope (SWR), as critical load-bearing components in engineering, require precise and real-time nondestructive testing of surface defects to ensure industrial safety. However, in practical engineering environments, machine vision-based methods face severe challenges such as complex backgrounds, varying lighting conditions, and low-contrast defect features, making it difficult to achieve both high accuracy and high real-time performance. To address this, this paper proposes a computer vision-aided lightweight attention-augmented approach for nondestructive steel wire rope defect detection. The core of the work is a comprehensive vision measurement system, featuring a custom multi-sensor synchronized device for full-coverage imaging and a standardized preprocessing pipeline to enhance defect features. Algorithmically, this paper designs the YOLO-CPC framework, which incorporates lightweight PartialConv-based modules (C3k2-PC1, C3k2-PC2) for efficient feature encoding and a channel-spatial attention block (CASAB) with SiLU activation for improved perception of low-contrast defects. Validated on a factory-collected dataset, the method achieves a state-of-the-art mAP@0.5 of 95.7% with only 5.7 GFLOPs. The method successfully balances computational efficiency with high measurement accuracy, presenting an integrated solution for the nondestructive inspection and quantitative assessment of surface defects in SWR.