a naive example of LivePortrait infer by ncnn
1.Retargeting(lip open/eye open/head pose editing)
2.Source image driving by video/images
PS: infer time: 28s/driving image
op type avg time(ms) %
Conv3d(no simd/vulkan) 27168.23 96.24%
other op 1059.54 3.76%
total time: 28227.77
1.https://github.com/KwaiVGI/LivePortrait
2.https://github.com/Tencent/ncnn