search

NVlabs / few-shot-vid2vid

Pytorch implementation for few-shot photorealistic video-to-video translation.
Other
1.79k stars 276 forks source link

Youtube playlist 3 - Pose model #77

Closed kaswani29 closed 3 years ago

kaswani29 commented 3 years ago

Third youtube plaliyst link has more than 90% videos with multiple people in single frame will that affect the model performance or does it help model generalize ?

tcwang0509 commented 3 years ago

Using the preprocessing code, it should crop only one person out at once.