I am trying to use face alignment network as backbone for faster rcnn pytorch pretrained model. I am using the net_s3fd structure as a feature extraction network and this is how I implement it.
`def create_model(num_classes):
The problem is when I start training on google colab or kaggle I get "Cuda out of memory" error. Is there a way to use this network as the feature extraction network with the pretrained weights other than the one I'm using?
I am trying to use face alignment network as backbone for faster rcnn pytorch pretrained model. I am using the net_s3fd structure as a feature extraction network and this is how I implement it. `def create_model(num_classes):
The problem is when I start training on google colab or kaggle I get "Cuda out of memory" error. Is there a way to use this network as the feature extraction network with the pretrained weights other than the one I'm using?
Thanks