cxgincsu/SemanticGuidedHumanMatting

Issues

• Paper claims SGHM to be 1.7x faster than MODNet, but testing locally SGHM is 2x slower than MODNet

  #20 opened 6 months ago by ani-mal
  0

• How to get same performance in image and video

  #13 opened a year ago by ToniButland1998
  1

• The difference between pred_alpha and pred_segment？

  #14 opened a year ago by goldwater668
  1

• fix cpu inference

  #15 opened 7 months ago by simplew2011
  3

• Np.int -> int

  #18 opened 7 months ago by sklipnoty
  1

• Training code

  #19 opened 10 months ago by LiYufengzz
  0

• Training code

  #16 opened a year ago by VladMVLX
  2

• Mobilenetv2 pretrained weights and training code

  #2 opened 2 years ago by arunsank
  11

• Use own rough mask?

  #17 opened a year ago by antithing
  0

• 训练代码能开源么

  #12 opened 2 years ago by Yuziyi1117
  1

• Pretrained model of the segmentation part

  #11 opened 2 years ago by ch1998
  2

• load pretrained-weight error

  #10 opened 2 years ago by xingha
  3

• 1.No training code found, will publish;2. about Video infer , Is there a time consistent design,Will flicker occur in video matting

  #6 opened 2 years ago by zhanghongyong123456
  1

• whether the dataset RWCSM-289 will be released

  #8 opened 2 years ago by kyrie111
  1

• about paper's gl calculate?

  #9 opened 2 years ago by zhanghongyong123456
  3

• Retain the human texture

  #4 opened 2 years ago by jerrinbright
  1

• Large image edge optimization is not very good

  #5 opened 2 years ago by z411500976
  2

• No module named 'inference'

  #3 opened 2 years ago by LiNO3Dy
  1

• Pretrain availability

  #1 opened 2 years ago by VolodymyrAhafonov
  1