You can't really do video detection it in real time and properly, without overheating the cheap device. On the SaaS we did we use a recorded segment of video so we can analyse it afterwards, check the colour changes in the face of the subject and try estimate heartbeat with loose accuracy. More like high, Normal, Low after taking a few seconds as base.
The idea of the ADHD device was just a cheap hack done in a day for myself.
Not a product or commercial as I don't feel claiming it can do to others what I looks like does to me.
Happy to share here FREE ; )
Greedo shouldn’t have asked about Han’s mother!
We try estimate a "confidence" level when saying something, but NOT able to guess if a lie. An actor can easily have full confidence in a script, but still be a lie.
We intended for roles that are client facing and where personal skills are key. Things like retail, costumer service, sales.... not technically or education based, temporary work and alike. Not that Orwellian yet ; )
Example: You need 40 people for an event to work a few hours as guides or hosts, from 100 applying the system can suggest a score and list them all in order so you can start watching or calling for interviews from somewhere.
There was research some time ago (jeez, 10 years ago...) [0] [1] that used video and... computer vision algorithms to detect a person's heart rate without any devices. Wondering if it could be used here i.e. use your webcam instead of an external hardware sensor?
Other thing - would you be open to publishing/sharing your algorithm / esp32 code (on Github or as a blog article)?
[0] https://news.mit.edu/2012/amplifying-invisible-video-0622
[1] https://people.csail.mit.edu/mrub/vidmag/