search

Sense-X / UniFormer

[ICLR2022] official implementation of UniFormer
Apache License 2.0
807 stars 110 forks source link

The picture has to be pre-trained #123

Open Aaron198513 opened 8 months ago

Aaron198513 commented 8 months ago

The picture has to be pre-trained, how do I change the parameters of the 2D convolution into the parameters of the 3D convolution

Andy1621 commented 8 months ago

Please check https://github.com/Sense-X/UniFormer/blob/f92e423f7360b0026b83362311a4d85e448264d7/video_classification/slowfast/models/uniformer.py#L387-L421