Question

Facial Recognition Input Stream Ideas

Answer 1

There are open-source codes and tutorials and how to turn your doorbell camera into a facial-recognition system that can notify you when a person has come, who the person is, and how many times they came.

Now then...we got a camera, we got code, and we can differentiate people.

Now then, mount it on glasses.

Answer 2

Here's the example of home-made doorbell face-recognition: https://medium.com/@ageitgey/build-a-hardware-based-face-recognition-system-for-150-with-the-nvidia-jetson-nano-and-python-a25cb8c891fd

Use it to
1. Find Parking Spaces: https://medium.com/@ageitgey/snagging-parking-spaces-with-mask-r-cnn-and-python-955f2231c400
2. Spy on people when they're...oh I shouldn't be promoting illegal activities here hehe
3. Find people that you know in a crowded area (HEY PROFESSOR WHAT ARE YOU DOING HERE AT MACYS?)

What else...detect cop cars when driving?
These ideas are revolutionary

Answer 3

This could also potentially be an all-reaching skynet used by the government if it can connect to the internet...hmmm

Answer 4

I feel like the police has already been developing this technology a few years ago thanks to some engineer that sold out

Answer 5

But it's a genius idea to have car-plate recognition systems along with a database of wanted cars lol.

Answer 6

lol PERFECTION

Answer 7

It is actually a third party app on the Google Glass,

https://www.google.com/amp/s/techcrunch.com/2013/12/18/google-glass-facial-recognition-app-unauthorized/amp/

Answer 8

Screw google, let's make our own backyard version.

Answer 9

You guys got to try this out on yourselves
https://www.betafaceapi.com/demo.html#page-header
That's what I'm expecting.

Answer 10

Rate how attractive people are based on a computer algorithm:
http://selfie.ai/

Answer 11

Good to know I mathematically rate 4/10

Answer 12

No duh, try a fashion model for Macys or something and you'll get a 10.
You need all the photo-shoppy look, symmetry, bright facial features, dazzling color combo...idk what features they're looking for but I'd guess it's mostly proportions of biometrics.

Answer 13

I wonder how mathematics compares to what QC users think :thonk:

Answer 14

O yea, we can also make our own create a model for what we think is attractive. But I'm not there yet.

Answer 15

please excuse me while I hijack this post a little, lol

Answer 16

lol

Answer 17

LOL

Answer 18

We can start off by gathering a database of pictures, give them to people, have people rate the pictures' attractiveness on 1-10 scale, take the mean, and we'll worry about finding measurements later on.
We'll have a table that looks like this:



Turn it into a matrix and so linear algebra magic on it to find a line of best fit.
OOF but collecting the data will take sooooo long

Answer 19

I mean I have a tesla class GPU, so training the data wouldn't take to long, it's always gathering the data that takes the longest tho

Answer 20

https://www.nvidia.com/en-us/data-center/tesla-v100/

Answer 21

On an unrelated note, I've been able to use a car-detection python program on my live webcam. (code attached if u wanna try it yourself)

Answer 22

Apparently I'm 7/10, heavily penalized cause non-symmetrical, feelsbad.

Answer 23

ouch...

Answer 24

Lol

Answer 25

Mine

Answer 26

My bf's

Answer 27

;-;

Answer 28

I'm confused where so you see stats?

Answer 29

Under classifier and measurements

Answer 30

IM CLASSIFIED UwU

Answer 31

Guess I look 15 lol

Answer 32

Lol, I like how it says my age 16 when I'm 17 and my bf's 15 when he's 16