This article proposes a new method for semiautomatically generating the image Mask required for training Mask R-CNN. Since manual labeling is very time-consuming and laborious to obtain the image mask, we propose a very simple and fast method based on graph cut to obtain image Mask method, which uses graph cut-based image segmentation to output pixel-level segmentation results and obtain Mask of the input image through image transform, and then, the Mask R-CNN-based surface defect detection method is implemented, which includes three different branches, namely, the boundary box regression and positioning branch, the boundary box classification branch, and the segmentation branch, to realize the function of locating, classifying, and segmenting defects at the same time. The experimental results confirm the effectiveness of our proposed method; under the premise of ensuring the detection accuracy of the Mask R-CNN method, the Mask required for training Mask R-CNN can be quickly and simply obtained.

中文翻译：

---

半自动掩模生成，用于电子元件检查

本文提出了一种新的半自动生成训练Mask R-CNN所需的图像Mask的方法。由于手动标记获取图像蒙版非常耗时且费力，因此我们提出了一种基于图割的非常简单快捷的方法来获取图像蒙版方法，该方法使用基于图割的图像分割来输出像素级分割结果，通过图像变换获得输入图像的Mask，然后实现基于Mask R-CNN的表面缺陷检测方法，该方法包括三个不同的分支，即边界框回归定位分支，边界框分类分支，分割分支，实现了缺陷的定位，分类和分割的功能。实验结果证实了该方法的有效性。