Define a complete set of facial features to work on for text-to-face generation

[DarkGeekMS]

The team needs to gather and decide the features that will be used for text-to-face generation, in order to :

• Fit the latent space axes on those features.
• Add those features to the multi-label text classifier.

NOTE : Only visual facial attributes need to be specified, however sparse facial features related to face landmarks are considered in face refinement. Moreover, sparse facial features already have some defined axes.

NOTE : Check the initial set at https://github.com/DarkGeekMS/portrait-synthesis-from-speech/wiki/Complete-Set-of-Considered-Facial-Attributes.