The model is very bad after the camera takes a photo

I tested your data. There could be more possible issues with it.

Some of the images are unfocused: 

In this part is not good image overlap (your back):

You are using featureless clothes:

As you can see most of features are found just on the carpet.

I wasn’t using the control point but the resulted model looks slightly better than yours:

Also it could help using more cameras, but I understand that could be hard.