Skip to content

Latest commit

 

History

History
19 lines (15 loc) · 797 Bytes

README.md

File metadata and controls

19 lines (15 loc) · 797 Bytes
Raw

Clothing Masking (Binary Silhouettes)

Tools to generate better fitting and proper binary masks/silhouettes from clothing images.

1) Masking with GrabCut using priors

Steps of implementation:

  • Binary thresholding (to get the primary silhouette/mask)
  • Filling operation (Floodfilling, to fill the big holes in the mask)
  • Opening operation (for salt noise removal)
  • GrabCut priors defining (absolute_foreground, probable_foreground, absolute_background, probable_background)
  • GrabCut masking

2) Masking with binary thresholding

Steps of implementation:

  • Binary thresholding (to get the primary silhouette/mask)
  • Filling operation (Floodfilling, to fill the big holes in the mask)
  • Opening operation (for salt noise removal)
  • Erosion operation (for thinning out boundary)