search

zhanglonghao1992 / One-Shot_Free-View_Neural_Talking_Head_Synthesis

Pytorch implementation of paper "One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing"
Other
720 stars 140 forks source link

有没有什么加速推理的方法或者可以借鉴的方案 #79

Open liuxuan111 opened 2 months ago

liuxuan111 commented 2 months ago

本身模型推理一个320帧视频大约花费40s，加上头部贴会原图5s，高清化（gpen）55s，耗时太长了，有没有什么优化方案可以参考

johndpope commented 2 months ago
Screenshot 2024-05-07 at 9 34 15 pm

This is 2 years old - more downstream work here https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=One-Shot+Free+View+Neural+Talking+Head+Synthesis&btnG=

madgnot commented 2 days ago

没有滚