About the output of inference.py

[lyx0208]

I try to test the try-on result with the inference.py according to the steps in the readme file:

1. download the checkpoints
2. download the caffe models
3. get the human parsing
4. put the photos into different paths
5. run inference

But the try-on output I get is rather poor than the figures in the paper: the margin is blurry and there's some white part at the margin. So is there anything wrong when I use your code?

[levindabhi]

Steps you mentioned is correct for inferencing image. I tried to match with paper details but I am also not satisfied with results I am getting on test data. There can be few reasons but I am not aware of the real cause.