Uses data from an infrared headset in combination with a machine learning algorithm to detect when the user is lying with ~80% accuracy.

When someone makes up a story, fabricates details, etc., much more of their prefrontal cortex is activated than when they tell a known truth. fNIRS is a strategy for determining brain activity by measuring scattered infrared light. The Brite23 headset (Artinis) uses fNIRS to measure activity in the prefrontal cortex. Thus the Brite23 is capable of detecting deception.

This project shows the robust accuracy (~80%) of the system when combined with 0.1-0.6 (butter) bandpass filtering and quadratic discriminant analysis. This project is really impressive because the conditions for collecting data at Cal Hacks are definitely less than controlled. This hack shows that infrared could be used for a street-ready lie detector in law and gov't.

Built With

Share this project: