search

ControlNet / MARLIN

[CVPR] MARLIN: Masked Autoencoder for facial video Representation LearnINg
https://openaccess.thecvf.com/content/CVPR2023/html/Cai_MARLIN_Masked_Autoencoder_for_Facial_Video_Representation_LearnINg_CVPR_2023_paper
Other
222 stars 20 forks source link

prediction model for custom video #15

Closed pradeepkc11 closed 6 months ago

pradeepkc11 commented 10 months ago

can you please upload the script for prediction model for single video so that i can check how it will perform

ControlNet commented 10 months ago

I think inference (feature extraction) can be done by simply apply features = model.extract_video("path/to/video.mp4") as shown in readme. If you want an example to extract features for a dataset. Please refer to the CelebV-HQ feature extraction scripts. https://github.com/ControlNet/MARLIN/blob/2547c63b611bcf10c0c46506563a618a72468441/preprocess/celebvhq_extract.py#L1-L43