Minority Mask Regions

Documentation

• Class name: SaltMaskMinorityRegion
• Category: SALT/Masking/Filter
• Output node: False

This node is designed to identify and isolate minority regions within a given set of masks, applying a threshold to distinguish these areas. It focuses on highlighting less prevalent features in the mask data, potentially for further analysis or processing.

Input types

Required

• masks
  • The masks input represents the collection of mask images to be processed, aiming to identify minority regions within each mask.
  • Comfy dtype: MASK
  • Python dtype: torch.Tensor
• threshold
  • The threshold parameter determines the sensitivity of the minority region detection, influencing which areas are considered minority based on their pixel intensity.
  • Comfy dtype: INT
  • Python dtype: int

Output types

• MASKS
  • Comfy dtype: MASK
  • The output is a tensor of masks, each corresponding to the isolated minority region within the original input masks.
  • Python dtype: torch.Tensor

Usage tips

• Infra type: GPU
• Common nodes: unknown

Source code

class SaltMaskMinorityRegion:
    @classmethod
    def INPUT_TYPES(cls):
        return {
                    "required": {
                        "masks": ("MASK",),
                        "threshold": ("INT", {"default":128, "min":0, "max":255, "step":1}),
                    }
                }

    CATEGORY = f"{NAME}/Masking/Filter"

    RETURN_TYPES = ("MASK",)
    RETURN_NAMES = ("MASKS",)

    FUNCTION = "minority_region"

    def minority_region(self, masks, threshold=128):
        if not isinstance(threshold, list):
            threshold = [threshold]
        regions = []
        for i, mask in enumerate(masks):
            pil_image = mask2pil(mask.unsqueeze(0))
            region_mask = MaskFilters.minority_region(pil_image, int(threshold[i if i < len(threshold) else -1]))
            region_tensor = pil2mask(region_mask)
            regions.append(region_tensor)
        regions_tensor = torch.cat(regions, dim=0)
        return (regions_tensor,)