Bluff Catcher

Inspiration

Poker is a game of skill, but reading people is a superpower. Professionals can catch subtle tells—blinks, lip biting, fidgeting, facial tension—but what if AI could do it too? Bluff Catcher was born from the idea of using computer vision to level the playing field and give players an edge in reading opponents, both live and on replay.

What it does

Bluff Catcher uses your webcam or uploaded video to detect potential bluffing behavior in real time. It tracks: Blink rate, lip biting, fidgeting/head movement, facial emotion (fear, anger, surprise, etc.) These metrics are fed into a scoring system that produces a live “bluff score” from 0 to 1, showing how likely someone is bluffing based on behavioural tells.

How we built it

I used MediaPipe and OpenCV to detect facial landmarks and motion. DeepFace was used to classify facial emotions. I built custom logic to track blinks, lip biting, and fidgeting, and developed a scoring system that combines these signals into a single bluff score between 0 and 1. The app features both a live webcam mode and a video replay mode.

Built With

• python