Facial coding - How it works

The science bit

Affdex qualitative facial coding software has been developed around robust and rigorous science and machine learning algorithms.  

Most facial coding systems rely on ‘posed’ data sets for training and testing their emotion classifiers.  Studies have shown that when subjects pose, their muscles work differently to an emotion display that is natural.

Affdex software is developed using classifiers trained and tested in spontaneous, real-world environments, presenting widely variable conditions to ensure the best possible accuracy when running live studies. 

Affdex emotion metrics are based on a three-step process:

  1.  The face tracker locks on to key features of the face.  These feature points known as ‘landmarks’ are used to identify the regions of the face that will be analyzed by the classifier.
  2.  Once the regions are identified, each pixel in that region is evaluated for shape, movement and texture by the classifiers. 
  3. The classifier then returns likelihood that the facial frame under analysis is representative of the emotional state. 

This is completed for each participant at high speed, then aggregated across participants and married to cognitive survey data to support more complex analysis.


Watch our Facial Coding webinar


Bob Granito, Interactive Video Productions
T.   +1 732.970.9446.

Related pages


Follow us on social media




  • BDI Acquisition

    June 22, 2017 - Schlesinger is excited to announce the acquisition of BDI Research, a research services company with facilities in Barcelona and Madrid, Spain.

  • 360 immersive video

    October 14, 2016 - Introducing 360° immersive video for focus groups. The system features 360° cameras that put observers at the center of the conversation, creating a whole new experience for observers and compelling video for researchers.

  • Young Professionals Grant

    September 5, 2016 - Schlesinger has once again collaborated with the QRCA to fund the Young Professionals Grant to give young qualitative research professionals the opportunity to attend the QRCA annual conference.

Copyright 2017 Schlesinger Associates