Readability and efficiency improvement for inference.

Weizhi-Zhong / IP_LAP

CVPR2023 talking face implementation for Identity-Preserving Talking Face Generation With Landmark and Appearance Priors

Apache License 2.0

679 stars 76 forks source link

Open IanLynnEE opened 9 months ago

IanLynnEE commented 9 months ago

Add inference.py and util, reconstructed from inference_single.py, for better readability and efficiency:

Change parameters of video output (cv2.VideoWriter) for better quality and efficiency.
Minor changes for efficiency, such as how and when to create tensors.