Overview
A computer-vision service that reads a desktop poker table in real time, applies a behavior library and records statistics across 100,000+ hands.
Problem
Table behavior and outcomes had to be observed from the screen and accumulated at scale for simulation and testing.
My Role
Requirements, architecture and development of the recognition loop, behavior library and statistics.
What I Built
- Real-time computer-vision recognition of the desktop table.
- A behavior library and statistics over 100,000+ hands.
Architecture & decisions
Classical computer vision and simulation. This is not a language-model or agent product.
Technologies
- OpenCV
Outcome
Delivered a simulation and testing service covering 100,000+ hands.
Case boundary
The client and product brand are omitted.