Blink Rate

Measures blink rate in a novel and patent protected way! This version was created using Unity and OpenCVSharp, an open source computer vision library for Unity. The first step in a remedial ocular correction system.

As a Patent inventor I was responsible for creating a Unity prototype of the proposed algorithm from an earlier academic prototype made in Matlab. I implemented a permutation testing system to find the best settings and hone in on unneeded variables for the user calibration process. I did extensive testing and quality passes while transcoding Matlab code into a Unity runnable C# version. The prototype existed in two variants: one for live webcam, and one for existing videos. This allowed us to run tests both in batches and on the spot for specific users in a live-testing environment.

The backend steps of the computer vision algorithm running on live webcam data. Accuracy should be around 85% and it runs at between 2ms and 4ms!
Slideshow of the detection steps. Using OpenCV we take webcam input and turn it into a positive or negative iris confirmation over time.