Cheaper, safer Smart Cities
In a list of the top 10 most surveilled cities in the world, eight were in China, with the other two being London and Atlanta.
Interestingly, London was in 6th position, followed closely by Wuhan, China.
In that same report, they look at the correlation between the number of security cameras per person, and the related crime and safety indices, concluding that there was very little correlation between safety and the number of CCTV cameras in the city.
We found little correlation between the number of public CCTV cameras and crime or safety
Smart City Surveillance a Question of Quantity vs Quality
Like many things in life, just because you have a lot of something, doesn't mean it's doing any good, and the same is true for surveillance systems in smart cities.
If you look around your city, you will probably spot CCTV cameras all over the place. Some will be highway cameras that the police control, others will be run by the local council, some will be privately owned on commercial or retail real estate. Then, of course, you have private homes with cameras from the sophisticated Ring doorbell, old style analogue cameras and perhaps something a little bit more esoteric like the home security camera system I custom built for my home.
Beyond the fact that they all record images, the common theme between all of these situations is that there is no common ownership or access, and that's both a benefit and a disadvantage.
You most likely don't feel comfortable with freely sharing the comings and goings through your front door with the police. But you might be more willing to do that if your home was burgled. What if your neighbour suffered a break in, or a car was stollen across the street, would you be more open to it then? Most likely yes. In fact, this is already being done in projects like Detroit's Project Green Light.
The arguments are mostly the same for all of these other types of security camera installation. Most of the time the camera owners would prefer the video feeds remain confidential, except for in certain situations, for particular use cases, as long as privacy is appropriately protected.
Without suitable frameworks in place to enable sharing of this camera infrastructure, everyone deploys their own system, keeping the feeds for themselves, multiplying the number of cameras, but not multiplying the benefit. Perhaps this is underlying the seeming lack of correlation between safety and the number of CCTV cameras in a city?
Adding up the costs.
Connecting all these cameras is an expensive business, storing and analysing all the data they produce is even more.
I heard that one of the most forward thinking UK Smart Cities, Bristol, budgets between £5,000 to £15,000 to connect each and every camera. That's a huge amount of money, especially if there is a camera in a similar location that they could perhaps share.
If you plan on storing and processing all the camera footage in the Cloud, then that has a cost too, another source suggested that can cost between £100 to £17,000 per camera, per month. (What Examples)
And then there's privacy to consider.
Do you remember when airports introduced the full body scanners?
There was a lot of panic and concern about private-parts privacy.
How the airports and manufacturers responded was to show you what they could see, and it wasn't at all what most feared. As my little illustration here shows, the systems would highlight the area for the staff to check. In my illustration below, apart from a snazzy pair of shoes it looks like the passenger forgot to take their watch off.
The lesson we should learn from this for smart cities is to be open with the citizens, show them what is being collected and how it is being used.
Protect Privacy, Reduce Costs
There are many valid, valuable benefits for collecting and using video information in a smart city. These benefits can briefly be summarised as:
- Improving efficiency of services
- Improved traffic management and mobility through the city
- Reducing pollution, protecting health and improving sustainability
- Economic growth - traffic delays reduce productivity, and smart cities attract new business
- and of course, safety and security.
Most of these involve improving efficiency in one way or another, e.g. efficiency of policing, efficiency of moving items and vehicles, all of those have a financial element to them, most have a privacy factor too.
There's a lot of public concern about privacy and facial recognition, so whatever we want to do with cameras and AI needs to take privacy concerns very seriously.
In order to live up to the benefits of a Smart City, we've got to find a way to more efficiency collect the information we need, in a way that safeguards the privacy and security of the citizens, and this is where the company I spoke with has an edge.
Privacy-Preserving Video Analytics (Computer Vision) on the Edge
With CCTV systems as they are right now, most record video and send it somewhere, typically the Cloud. This means you need to have good bandwidth, lots of storage and some pretty robust security and privacy policies in place.
The irony of this, is that most of the data you end up paying to transfer, store and protect is of little value. But there's another way.
What about giving the cameras some autonomy and intelligence to sift through the reams of data and only send the things to the network that really matter? Sounds good. That's what you can do when you apply AI to the Edge (What does The Edge mean?), and this is what NATIX is doing.
Working with chip manufacturers (e.g NVIDIA and NXP) and camera manufacturers (e.g. BASLER), NATIX's software stack equips cameras with artificial intelligence, so they can look for the events that matter at the edge, quickly and efficiently reporting just the events and data that are important.
Those events could be large scale events like crowds of people gathering, congestion or car park availability. On the smaller, more individual level, the events a city might be interested in could include unauthorised people in restricted places, cars jumping red lights, unexpected objects in unexpected places (e.g. suspicious packages), citizens at risk of harm, obstructions on the roads or road accidents.
Things get really clever when you start to combine multiple cameras in the detection and tracking of things. Imagine a scenario where a camera on a bus sent an alert that there was a bag left unattended an a library. This is how multiple cameras, collaborative equipped with AI at the edge could handle that situation:
- The library camera spots solitary bag, and sends a warning message to security about a potential threat.
- The security team make their way to the location, whilst ....
- The library AI Edge processor searches back through the locally stored video to identify the person who left it. Then collaborates with other locally connected cameras AI Edge processors to discover that the individual left the build a few minutes ago, but the street camera can now see them returning. The individual returns, retrieves their accidentally forgotten bag and leaves the building.
- The system determines the threat is removed, and sends a message to the security team to back down.
AI, Fog and Edge working in a real-world exampleA project NATIX are working to help with COVID-19 exit strategies
The whole world is trying to figure out how to return to some kind of normality, and some countries are further ahead than others. One thing seems to be sure, that is for a foreseeable future face masks are going to be a more common requirement in public places.
NATIX is currently working to help analyse occupancy levels in shopping malls, and the ratio of people in that building wearing/not wearing facemarks.
The system uses multiple camera plus AI Edge processors to keep track of the total occupancy, counting people entering and exiting the building through multiple doors, each camera also looks for whether the individual is wearing a mask.
The system then publishes data about the occupancy levels and the ratio of masks being worn. This information is very valuable for the building operators, and compared to having multiple camera streams being monitored by a security person, much more cost effective and accurate.
This use case extends beyond shops, and could include public spaces, airports, and offices.
Computer Vision Demo
Watch the video of the privacy protecting face mask detection system in action click the menu button here.
This Could Make Any City SmarterUpgrade and retrofit existing cameras to make them smarter, more cost effective
As mentioned at the top of this article, in some cities there is a high degree of redundancy in the camera infrastructure. This is due to the ownership models and the restrictions on sharing them.
Technologies like this from NATIX mean that cities can upgrade existing camera infrastructure, deploying advanced artificial intelligence models to the edge of the network, making even older analog cameras smart. This enables cameras to do more through higher degrees of automation and intelligence, and facilitates sharing of the camera infrastructure across different entities, such as the council, business premises and police, without privacy or regulatory concerns.
And since all of this is done at the edge, the additional impact on the networking infrastructure or cloud computing costs is negligible, nothing like the eyewatering numbers mentioned above.