Software Describes Surveillance Footage In AI-Generated Text

holy_calamity writes "A computer vision research group at UCLA has put together a system that watches surveillance footage and generates a text description of the events in real time. It only works on traffic cameras for now but demonstrates how sophisticated computer vision is becoming. Interestingly, the system was built thanks to a database of millions of human-labeled images put together by Chinese workers."

Comments

Expectation of privacy

[Anonymous Coward]

There needs to be an expectation of privacy regarding recordings of people in public places. There is a huge difference between being seen vs. having one's every public move recorded, indexed and archived.

Re:Expectation of privacy

[MyLongNickName]

Good luck with that. So, if I am in public I should expect that anything I do not be recorded, talked about or written about? I do not know how you expect to enforce that.

Re:Expectation of privacy

[IndustrialComplex]

Good luck with that. So, if I am in public I should expect that anything I do not be recorded, talked about or written about? I do not know how you expect to enforce that.

You should have some expectation of privacy because we need to have SOME privacy in order to function as human beings. Generally the expectation goes back to what you would feel comfortable with if it were performed by a physical person. And I'm certain that if it were somehow possible to assign a person to follow and document every move, and action for every person in the US we might have a slight problem with that.

We run into a hell of a lot of trouble when we allow our standard definition of privacy which involved 1800s methods to be applied to our current level of technology.

The basic problem is this:

As technology improves, our expectation for privacy decreases. So using expectation of privacy as the measure for what should or should not be private is a HORRIBLE practice. It essentially means that as a technology or practice becomes ubiquitous, it becomes acceptable.

Since we have no means to resist an application of technology*, I urge everyone to dump this 'yardstick'.

*In practice, you do not get to opt-in or opt-out of having a privacy invading practice applied to you. It IS applied, and then you have the option to petition against it's application. Often, you don't even know that your privacy is being violated for years. As a result, these practices become common before the first complaint can even reasonably be raised. Even then, this ignores the issue of having previous complaints dismissed by judges who are ignorant in the field of the technology being discussed.

Re:Expectation of privacy

[drsmithy]

There needs to be an expectation of privacy regarding recordings of people in public places. There is a huge difference between being seen vs. having one's every public move recorded, indexed and archived.

The word you're after here is anonymity, not privacy.

Re:Scary

[mystik]

What's broken then?

The Laws?

Or the Enforcement?

Re:Scary

[ElectricTurtle]

Yeah, too bad that's a gross generalization that doesn't correlate with reality. Besides the fact that the concept of wealth inequality as moral negative is nonsense, it doesn't take too much analysis to see that while the US and Mexico may have similar ratios of rich to poor (which by itself is misleading, as 10^4:10^3 is the same ratio as 10^2:10, but the magnitude is different, so the case really is that the poor in the US are richer than the poor in Mexico, and the rich in Mexico are poorer than the rich in the US. The ratio ultimately is the same, but the magnitude is different, which is expressed in the difference in the quality of life), crime in Mexico is worse. Similarly, in 'more equal' countries according to your favored methodology like Columbia, Nigeria, etc. crime and quality of life is worse than in 'less equal' places such as Hong Kong. Your theory simply does not correlate to reality, but I doubt this will stop you.