Your take on this is inaccurate, and this is not “facial recognition”. This is pattern recognition, and shoddy at best from the description. It’s using converted 2D imaging data with heat sensor data, and reduxing the resolution down to another 2D map, and ATTEMPTING to make a match. This is what we would call “fuzzy” recognition, and is not accurate at all. Fuzzy inference takes any range of accuracy as a success, meaning if it gets it right 5% of the time, it’s considered accurate.

Bottom of the barrel bullshit that doesn’t work. Don’t even take my word for it, you can look up this same pipeline model of inference in countless GitHub repos that the original work came from and understand why.

Here’s an example:

• Person 1 goes for a run, then turns direct to FLR camera
• Person 2 who looks similar, but 50lbs heavier walks out of their house and walks towards camera
• 15% match on inference

The military would consider this a detection.

YOU on the other hand are talking some crazy conspiracy theory noise about how these cameras are used in crowds and can differentiate using heat detection maps on people wearing face paint.

NOT HOW THAT WORKS

This is saying they can map thermal to known faces. But this is far from being able to read faces through materials. Thermal/infrared doesn’t provide that kind functionality, at least to my knowledge the signal would bleed into the mask. It could work for glasses or maybe thin masks.