The following link is to a Youtube channel called SmarterEveryDay. The person running the channel has an engineering background but I haven't watched enough of his videos to know anything more than that.
I've posted this because it's about object recognition by computer surveillance. This is not a new concept but the video shows new ways it could be used.
This fascinated me not because it was about the ability for video surveillance to detect firearms in active use but because it highlights how far computer & video technology has come since DC was released. It also fascinates me because of the potential for detecting other objects that the observer may or may not find desirable.
So talking about how it could be used in a game, this is not just about how the corps or the police could use this tech to prevent crimes but aslo about how corps could use it to highlight their own products in the hands of the general public (for use as advertising) or you could (if the video tech is good enough), have it used by criminal gangs to recognize expensive clothing items and thus they can locate more profitable targets for robbery.
This sort of software has been around for some time, it's been used for traffic management for such things as simply keeping track of the number of vehicles using a given intersection to keeping track of the number of a specific vehicle type (e.g. trucks as opposed to cars) using a particular road.
It's been used in the medical field, for example to check the number of motile sperm in samples from men who may have fertility problems.
The big difference with this particular application of the software, is that it checks the video feed for desirable/non-desirable objects and then sends a notification to specified people including an image of the object in question.
It does have some limitations at the moment, mostly due to two potential issues.
First is that the software requires many thousands of images of both desirable and non-desirable objects to be scanned and highlighted so that the software can "learn" what is a positive ident and what is a non-problem object but also to capable of correcting a false-alarm.
Second is video technology not being able to provide enough pixels for the software to recognize a desirable/non-desirable object past a certain range from the camera. So while that limits the abiliity of the detection attempt, it's not likely to be a problem in areas where it would prove most useful, e.g. controlled access points to buildings, the main floor of a bank and so on.
However, if coupled with a high quality video device, there's no reason why this software could not be applied to suburban or city streetscapes, carparks, plazas, outdoor tourist spots and so on.
I can even imagine the software being commercialized for the general public e.g. trophy hunters using something like this to check if the deer they're after has "prize" quality antlers while the hunter stays hunkered down in their shooting hide.
So, to the video...
https://www.youtube.com/watch?v=Lh0x54GC1sw
