On the issue of semantic mask segmentation in data processing.

[Garfield9986]

Dear author, I sincerely admire you for this excellent work, I am new to this field, and I am very impressed to see your work. There are a few things I don't understand about your work, I see that when data is processed, masks of objects separated by different semantics are generated under the segmentations folder, and I don't know where in the code this step is performed. At the same time, I would like to ask you, if I want to generate a corresponding Gaussian mask based on the semantics of an object on an image (such as the content of the third image in your article), where should I call the interface? I would appreciate it if you could answer my questions!