I was recently introduced to eigenvectors. With your dataset, wouldn't it be possible to find the one official with the "most average-looking" face by comparing each recorded face to the dataset's eigenface (that is, the average'd version of all faces in the dataset)?
![image showing The Average Face of Chinese Bureacracy [OC]](d55cc98e-f3b5-5860-8aad-71db8a608496.jpg)
everest4ever on October 22nd, 2017 at 16:24 UTC »
The average face of 1014 Chinese officials in central government and high-level local government (provincial administrators).
Portraits: People's Daily online -- database of Chinese officials (人民网中国政要资料库)
Facial landmark detection: Matlab Central file by Silvia et al. (2012): Detection of Facial Landmarks Using Local-Based Information
Data analysis and plotting: Matlab
iamjusthonest on October 22nd, 2017 at 17:19 UTC »
The secret to bureaucracy is to blend in and disappear among the crowd while collecting tax payer cheese.
Scarbane on October 22nd, 2017 at 17:24 UTC »
I was recently introduced to eigenvectors. With your dataset, wouldn't it be possible to find the one official with the "most average-looking" face by comparing each recorded face to the dataset's eigenface (that is, the average'd version of all faces in the dataset)?