Steel surface defect detection based on Improved YOLOX Algorithm

Abstract As one of the very common industrial products, strip steel plays a important role in various fields. Steel production has received extensive attention. In production, scratches are unavoidable due to numerous forces beyond human control. Cracks and other flaws appear, their presence direct impact on product’s quality. Therefore, it is great significance develop efficient accurate surface defect detection. this study, most widely used metal material selected as research object, its detection problem studied accordingly. The activation function loss improved basis YOLOX model. Resolve overexposure content distortion image caused by high reflecting characteristics strip, which degrades quality, embedding sub-attention mechanism enhance regional hidden small targets. experimental results show that mAP model for surfaces 80.7%, 4.1% higher than original algorithm, can reduce false missed rate defects.