MODERN CAPTCHAS PROVE MASS SURVEILLANCE AND IMAGE RECOGNITION, feb 2019
H: not entirely sure why this popped into my head...
H: Okay, but why? zillions of different images of walkways, highways, residential streets, storefronts... and it's not like someone is sitting there putting descriptors on each of these images. It's gotta be automatic / image recognition that is assigning the different labels (car / bus / fire hydrant).
H: well, for instance, when you drive around freeways you always see cameras. Not just there, basically any stoplight has cameras on it these days. They just never really mention them or tell you about it. You can research CCTV (MUH CC 33!!!!) in london they are open about it. If you're out on the streets you're being recorded.
H: Image recognition is very good these days, so, all those cameras on the streets can recognize here's a human, here's a car, here's a stoplight, here's an airplane, here's a firehydrant, etc etc.
H: The captchas basically prove this, unless, you think they have an army of chinese people in a basement somewhere looking through tons of images of things and giving them labels of what they are. That makes no sense and uses way too much manpower. It's all automatic. They can probably search their system for 'red car' and be shown all the stoplights where their cameras are currently seeing a red car, how many red cars, etc.
P: good thing AIs are dumb though. people should learn this and exploit it.
H: walk around wearing a burka trashbag with a realistic looking firehydrant image on it
P: the latest fashion sensation
H: surprised those weird morph suits never got popular with famous people images pasted on them.
/end. original formatting was weird.
edit: removed tons of repeated comments. like i said, this one was a bit odd.
MODERN CAPTCHAS PROVE MASS SURVEILLANCE AND IMAGE RECOGNITION, feb 2019
-----------------------------------------------------------------------
H: not entirely sure why this popped into my head...
H: Okay, but why? zillions of different images of walkways, highways, residential streets, storefronts... and it's not like someone is sitting there putting descriptors on each of these images. It's gotta be automatic / image recognition that is assigning the different labels (car / bus / fire hydrant).
H: well, for instance, when you drive around freeways you always see cameras. Not just there, basically any stoplight has cameras on it these days. They just never really mention them or tell you about it. You can research CCTV (MUH CC 33!!!!) in london they are open about it. If you're out on the streets you're being recorded.
H: Image recognition is very good these days, so, all those cameras on the streets can recognize here's a human, here's a car, here's a stoplight, here's an airplane, here's a firehydrant, etc etc.
H: The captchas basically prove this, unless, you think they have an army of chinese people in a basement somewhere looking through tons of images of things and giving them labels of what they are. That makes no sense and uses way too much manpower. It's all automatic. They can probably search their system for 'red car' and be shown all the stoplights where their cameras are currently seeing a red car, how many red cars, etc.
P: good thing AIs are dumb though. people should learn this and exploit it.
H: walk around wearing a burka trashbag with a realistic looking firehydrant image on it
P: the latest fashion sensation
H: surprised those weird morph suits never got popular with famous people images pasted on them.
/end. original formatting was weird.
edit: removed tons of repeated comments. like i said, this one was a bit odd.
(post is archived)