Table 3.

NME comparison for our transfer trained 2-input network and the current state-of-the-art. Best performance is highlighted in bold.

Methodology	Mean NME
Proposed 2-input Method	0.0074
3DDFA_2 [19]	0.0127
PRN [4]	0.0134
VRN_Guided  [16]	0.0187
3DDFA [18]	0.0227