Using Computation and Analysis for Business Insights
4
|Profile Picture for Company Leadership
Introduction
After picking a name for the new company, the next step might be to pick a good picture for the CEO or leadership of the company. You might want the CEO to project a youthful appearance or a slightly older and more experienced appearance. Can we use computation to estimate age across several photos to help select the best photo?
Predicting Age and Emotion for Various Photos
Predicting Age and Emotion for Various Photos
Wolfram Language includes commands that can create deep learning models or use existing deep learning models on your data. The function FacialFeatures includes a deep learning algorithm to predict the ages in a list of images.
The amount of styling and makeup in two photos of the same person can affect the age prediction. We can use Kenneth Lay as an example. First, let’s analyze the photo from Wikipedia from his arrest in 2004. After opening the webpage in a browser, you can right-click the image, choose “Copy image” and then paste the image into a Mathematica notebook:
In[]:=
FacialFeatures
Out[]=
Image
,Age72,GenderMale,Emotion
In[]:=
FacialFeatures
Out[]=
Image
,Age52,GenderMale,Emotion
The photos might have been taken at different times, but styling seems to have an effect on the estimated age for Kenneth Lay.
In the examples above, Wolfram Language is also using deep learning to predict Kenneth Lay’s emotion. It might be preferable to use photos that show happiness instead of a neutral expression:
In[]:=
FacialFeatures
Out[]=
Image,Image
,Age61,GenderMale,Emotion
,Age50,GenderMale,Emotion
It’s very straightforward to use the same Wolfram Language calculations on different images to experiment. An image can be dragged and dropped into a Wolfram Notebook to include it in calculations, like the inputs above.
Summary
Summary
Computers are likely not perfect at predicting age or emotion, but quantifying these characteristics can be a nice way to compare pictures in order to choose the one that presents the desired image for the CEO and the company.