Something interesting I want to find out, which I actually think is actionable at the moment.
I’ve thought. Am I in the background of other people’s photos?
I’m sure I am. You are out in a club and someone takes a photo, you are there. Or you are at a touristy place, someone takes a photo and you are there.
So I think I’ve derived a way to test this. Unfortunately I don’t have time atm, so I’m putting this out there. Whoever wants to do this, please do, because I’m really keen to see the results!
In order to increase the chance of detecting something, the sample size is reduced. Okay. So this is how you test it:
1) Grab your facebook profile. Get all photos tagged of you in it. Train your face with a facial recognition API (such as Lambda’s one).
2) Now get your friends of friends through the API (reducing sample size and increasing chance of detection). Friends of friends are more likely close by with which you haven’t coincidentally interacted it.
3) Now those friends of friends might have public photo albums of events. Dump all those photos.
4) Depending on location, you can filter out photos of places you know you haven’t been yet (such as public vacation albums).
5) Now run the model you trained on those public photos and see if your face pops up in the background somewhere.
I’m really curious how much data you’ll get and if you’ll be to find photos where you are in the background. If someone wants to do this, please do!