What you can see from the video below is definitely the most advanced real-time surveillance system I have ever seen.
It’s quite possibly the best object recognition system I’ve seen too.
I’ve dabbled in computer vision before, so I know first hand how difficult it is to get a computer to recognise objects in images when all it sees is monstrous array of numbers!
That’s what makes this system so impressive. It really does look like something from another world. Check it out:
So what are we looking at here?
Well, it’s basically CCTV footage of a road in China and at first glance, it appears to be carrying out surveillance on everybody that comes into view of the camera.
This amazing system is the work of a Chinese computer vision startup called SenseTime. The software is a visual scenario analytics system – basically a very sophisticated object detection system which uses deep neural networks (AI) to classify moving objects as either people, cars, buses or any other entity.
What makes this system remarkable is that it can even detect gender, clothing and the different types of vehicle that it’s looking at, all in real time.
You’ll also notice that it can even detect objects when they are occluding each other. For example when a person walks side-by-side with another person, the software still manages to distinguish them both and is able to draw bounding boxes around two people. This I think is absolutely amazing.
What does the text mean next to each person?
I can’t be certain, as I can’t read the language (maybe someone can help me out here?), but there’s a lot of speculation on Reddit that the text boxes are descriptors that are a result of the object classification algorithm. In other words, the algorithm is classifying each object as being a man, woman, adult etc. It also seems to pick up on items of clothing too and their color.
Here’s an English version of the same system, which is equally as impressive:
What is the system used for?
Well obviously it looks like a surveillance system, but the technology behind it is being applied in many areas.
The company behind it has been contracted to apply it’s technology to facial recognition within banks, so that people can make transactions, and open accounts using just their face.
The system is also able to read ID cards and compares the image on the card with an image taken by a smartphone camera. It’s not just a straightforward photograph however, it performs ‘liveness’ detection to ensure that it really is you taking the photo and not just some scammer holding up a photograph to game the system.
It is also being used by China’s biggest social network (Weibo) as a photo album feature.
The technology clearly has lot’s of potential. It could look for traffic patterns and adjust the timing of the lights appropriately, or adjust speed limits. It could automatically detect speeding vehicles or those that jump red lights.
It could also be used to count the volume of traffic passing through the junction every day.
My personal view is that we need this kind of system in public places, especially given the amount of terrorism that we see these days. It is physically impossible for the police and intelligence agencies to monitor terrorists and criminals without this sort of technology.
And I guess, if it’s only used within public places, then I’m OK with that. Others may not agree here, as it could be seen as an invasion of privacy.
The only potential problem that I see is that retailers could use it to spy on your buying habits and potentially sell the data to other companies.
Facebook especially would be interested in this type of technology (if it hasn’t already developed its own version).
What do you guys think? Are you afraid of this technology and what it could be used for, or all you all for it like myself? Let me know what you think in the comments.