A
AI BASED LIE DETECTOR
Unveiling The Truth
The problem AI BASED LIE DETECTOR solves
Detecting deception is critical in interviews, examinations, and security checks, but existing methods like polygraphs are intrusive, expensive, and unreliable. This project provides a non-intrusive, AI-based solution that analyzes speech features and facial cues to classify responses as Truth, Lie, or Uncertain in real time.
Challenges we ran into
Data Availability: Limited real-world datasets for lies vs truth.
Noise & Variability: Background noise and different accents affect accuracy.
Feature Reliability: Micro-expressions and voice stress can vary across individuals.
Ethical Concerns: Privacy, consent, and risk of misuse.
Model Generalization: Risk of overfitting on small or biased datasets.
Tracks Applied (1)
$300(Open): Cash Prize
Discussion
Builders also viewed
See more projects on Devfolio