search

tsing90 / pytorch_semantic_human_matting

This is an unofficial implementation of the paper "Semantic human matting":
https://arxiv.org/pdf/1809.01354.pdf
83 stars 18 forks source link

How about MobileNet as T-Net? #11

Closed pourfard closed 5 years ago

pourfard commented 5 years ago

I saw that lizhengwei1992 has used MobileNet as T-Net. PSPNet is much larger and is very slow. How do you think about using MobileNet?

tsing90 commented 5 years ago

I think it depends on your task. I tried MobileNet, for simple portrait images, it seems OK, but if your task requirement is higher, then you may have to consider a more powerful architecture.