利用卷积神经网络进行缺陷检测

Surface defect detection plays a key role in controlling the quality during manufacturing process for strip steel. However, traditional defect detection of strip steel in industry is still performed manually, which can’t meet the requirement of real-time inline detection due to yield low efficiency and high missing rate. So far, computer vision based techniques for defect detection has attracted much attention of researchers, which has important theoretical and practical value.