fungtion

Can add the support of proxy someday?

I found that there are 3 upsampling layers in the paper, while only 2 upsampling layer in this implementation. Could it be any different?

Hi, as your SIW network was trained on FFHQ, so if I test on my own real images, do I have to make alignment on those segmaps output from SOF?

Hi, I found it needs a inpainting model to complete the unvisible part of the face, could you share your pretrained inpainting model?

Are the released models same with the ones used on web demo?

1

Hi, I tested the onnx models in your repo and web demo, respectively, with the same images as follows, but got very different results: ``` res = 1024 im =...

Extracting face not working on GPU

2

When `5_data_dst_extract_faces_S3FD.sh` running on my ubuntu 22.04 with RTX2080, no face is detected, but it works fine when using CPU

I found a bug in [code](https://github.com/walnut-REE/sof/blob/main/dataset/face_dataset.py#L183)，and the "shuffle" should be set to "Fasle", it leads to the mismatch between image and the extrinsic parameter.

When create decoder using PyNvVideoCodec as follows: ``` nv_dec = nvc.CreateDecoder(gpuid=0, codec=nv_dmx.GetNvCodecId(), cudacontext=0, cudastream=0, usedevicememory=True) ``` it reported error: No AVFormatContext provided, I have no idea how to fix it