Applied 3D asset to live action shot

Customers are going to record themselves with a smartphone and we have to apply a 3D asset (helmet) to a live action shot. Maximum duration of each shot is 30 seconds and there are 150 videos to track the head movement and apply the 3D helmet. There will be some interaction between the subject and the helmet, such as closing the visor, and projecting graphics. Please take a look at the example.