Applying Facial Recognition to Criminal Justice

Team: 63

School: Portales High

Area of Science: Social Sciece


Interim: Team Number: 063
School Name: Portales High School
Area of Science: Social Science
Project Title: Applying Facial Recognition to Criminal Justice

Problem Definition:
Dishonesty is a factor of life; chances are any give person reading this has lied within the week, if not within the day. But how can humans sort fact from fiction, and is their a common thread between the behavior of liars? While lying is often done with good intentions, there are a number of other situations in which lying is ill-intentioned, such as a criminal making a false statement. In situations like these, having ways to spot deception can make a case easier. Psychologists, as a result, have been searching for “tip-offs” that a person may be lying, and the Facial Action Coding System, or FACS, was created as a catalogue as facial and bodily cues that may indicate deception from a human (Adelson). These small facial expressions are also known as “microexpressions”, and they are difficult to hide. Because these expressions last for as little as 1/25 of a second, it is very difficult for untrained eye to spot. So that leaves the problem of creating a code which can efficiently record and analyze a face to detect falsehoods.

Problem Solution:
The program we plan to draft through Python would be responsible for recognizing basic facial characteristics and identifying changes in the face that may indicate an ulterior meaning, a or falsehood being told. The code should be able to recognize physiological characteristics such as wrinkling around the eyes and mouth, as well as possibly provide a frame-by-frame analysis for humans to analyze. But ideally, our program should be able to identify these physiological changes on its own, as well as provide some sort of likelihood that the face being scanned is lying.

Progress to Date:
As of today, in addition to a semi-functional facial-detection program, our team members have come away with a knowledge of the Python syntax, semantics, and a knowledge of how to detect fabrications by identifying key facial characteristics. To identify key facial characteristics, we have utilized OpenCV. Currently, the program identifies the eye, and scans for aversive behavior relating to the eyes.

Expected Results:
After the careful designing and testing of our program, the system should ideally be able to accurately detect falsehoods using changes that may go undetected by the human eye alone. More advanced models may be able to check for falsehoods using characteristics mapped from the entirety of the face. Our technology may advance to a point in which we have developed a program that gives readings accurate enough to be admissible in a court of law. Using this technology, we will be able to accomplish a task that would be considered by many to be near impossible for artificial intelligence.


Team Members: Kerstiy Laman, Stephen Villanueva, and Bella Roy

Sponsoring Teacher: Jack Willis

Adelson, Rachel. “Detecting Deception.” American Psychological Association, 2004, www.apa.org/monitor/julaug04/detecting.aspx.

Barbara, Nigel. “Do Lie Detectors Work?” Psychology Today, 13 March. 2013, www.pshychologytoday.com/us/blog/the-human-beast/201303/do-lie-detectors-work.

Micro Expressions | Facial Expressions.” Paul Ekman Group, www.paulekman.com/resources/micro-expressions/.

Ried, John E. The Lie Detector in Court, 4 Depaul l Rev. 3, 1954.

Tiwari, Shantnu. “Face Recognition with Python, in Under 25 Lines of Code – Real Python.” Real Python, Real Python, 11 June 2018, realpython.com/face-recognition-with-python.


Team Members:

  Stephen Villanueva
  Kerstiy Laman
  Zara Roy

Sponsoring Teacher: Jack Willis

Mail the entire Team