“Computer Assisted System to Increase Speed and Reliability of Manual FACS Coding.”

RealleaR, LLC. & Naval Research Laboratory, 9/1/07 to 8/31/09.

Joint work with Fernando De la Torre.

FACS (Cohn, Ambadar, & Ekman, 2007; Ekman & Friesen, 1978; Ekman, Friesen, & Hager, 2002) coding is the state of the art in manual measurement of facial action. FACS coding, however, is labor intensive and difficult to standardize. A goal of automated FACS coding (Cohn & Kanade, 2007) is to eliminate the need for manual coding and realize automatic recognition and analysis of facial actions. Success of this effort depends on access to reliably coded corpora of FACS-coded images from well-chosen observational scenarios. Completing the necessary FACS coding for training and testing algorithms has been a rate-limiter. Manual FACS coding remains expensive and slow. FACS coding is slowed by the failure to apply current knowledge in computer vision to the task of computer-assisted coding. The speed, efficiency, and quality control of FACS coding can be increased dramatically by making use of new machine learning and computer vision algorithms to preprocess video streams for human coders and progressively decrease manual efforts. In preliminary testing, FastFACS has demonstrated concurrent validity for action unit onsets and offsets comparable to that of FACS coders and 50% or greater reductions in manual coding time. The larger goal of this effort is robust, fully automated FACS coding.