When using track membranes for the template synthesis of nanomaterials, it is important to have data on their characteristics (porosity, pore distribution on the surface, etc.). A method for the segmentation of track-membrane images to search for pores on the surface using the UNet++ model is implemented. The result of the program is a binary mask, consisting of black and white areas, highlighting the background and pores, respectively. Analysis of the masks has shown the possibility of obtaining such characteristics of track membranes as porosity, localization, and the number of pores.

中文翻译：

---

使用计算机视觉对轨道膜进行认证的方法

当使用轨道膜合成纳米材料的模板时，了解其特性（孔隙率、表面上的孔隙分布等）的数据非常重要。实现了一种利用UNet++模型对轨道膜图像进行分割以搜索表面孔隙的方法。该程序的结果是一个二进制蒙版，由黑色和白色区域组成，分别突出背景和毛孔。对掩模的分析表明，有可能获得轨道膜的孔隙率、定位和孔隙数量等特征。