Fast embeddings for head pose

[jakeelwes]

Hey just wondering if there was a simple way of getting fast embeddings just for the pose/coarse style (face landmarks & headPose). I was thinking of trying a realtime generator responding to movement through a webcam input - has anyone tried this & do you think it could be possible?

[pbaylies]

Hi @jakeelwes it's totally possible, @gradient-dissenter has done it. I'd suggest training a ResNet or similar to predict one or two of the coarse dlatents. I've got code in my repo for training ResNets and EfficientNets, although out of the box it tries to predict all the dlatents.

[jakeelwes]

Amazing, thanks @pbaylies I'll give it a go :)