search

ZHKKKe / MODNet

A Trimap-Free Portrait Matting Solution in Real Time [AAAI 2022]
Apache License 2.0
3.77k stars 633 forks source link

Perform well on synthetic images but bad on natural images. #149

Open syfbme opened 2 years ago

syfbme commented 2 years ago

I have done a transfer learning to matting certain objects. Below is my steps: 1: train modnet from scratch. train dataset contains around 60k (6k foreground images and 180 background images) 2: train modnet with soc(30k natural images without synthetic) The model(w/wo soc) performs well on synthetic images but bad on natural images. I think the S branch is is not good enough since the model can't locate the target object well... Maybe the synthetic way is not good enough(I just paste the foreground's target object to background image). Or maybe i should replace the S branch's backbone? Do you have any suggestions? Thanks a lot.

Tanlang180 commented 2 years ago

What's your dataset? And I would like to ask where to get the dataset from,haha