* fix benchmark

* add fomm and stylegan benchmark

* fix version of imageio

* add copyright of psgan, test=document_fix

* add copyright, test=document_fix

* add compress network of fom

* add training od first order motion
* Modify codes according to reviews
* modify single person and multi person process of fom model. 
* Git rid of sklearn, skimage

* refine code

* test

* fix apps

* update readme

* rm unused code

* fix apps output when input is image

* clean code

* update requirements.txt