argman

@louyanyang , what about your experiment result after more epoches? does num_deformable_group affect much ?

@louyanyang , got it, tks!

@YuwenXiong , the paper did not mention the deformable_group param, have you done any experiments of this?

@BowieHsu , thks! I will try, and will post my result here.

@BowieHsu , btw, can you share your trained model ? As i am using tf-1.3, so need to check whether some changes in tf.

@BowieHsu , after 6 hours of training using 4 gpus, the loss curve is 

@BowieHsu , thks for your model, i can get meaningful result now! The model is really hard to train..

@anewell , thanks for your detailed introduction, I have two more questions. 1. during training, if a joint is invisible, but have a coordinate, do you ignore this during training...

you can assign views to views, like ``` xt::view(channels, 0) = xt::view(....) ```

As i want to merge your code, I find there are many changes to the base class..