Visualization

[ZNorah]

Dear author, we can see that the visualization in your paper is very good, especially the visualization of the 3D surface data distribution. Can you ask how it is implemented?

[wy1iu]

Simply set the output feature dimension as 3. It corresponds to the weight matrix size of the layer before entering the classifier layer.

[CloudWalking0]

Simply set the output feature dimension as 3. It corresponds to the weight matrix size of the layer before entering the classifier layer.

Sorry to bother you. May I ask few questions about this?

1. Do I need to retrain a model with like 10 identities and set the feature dimension as 3? If so, what are the epoch and the learning rate?
2. Could you provide more details about drawing the 3d figure?

Thanks and have a nice day~