Rise of the Smart Cams
No, they’re already here. And get ready to see a lot more of them, as the hardware needed for a Smart Cam is becoming very affordable and its applications are near endless.
So, what’s a Smart Cam? It’s actually a normal ‘dumb’ IP or USB camera connected to a computer which runs software to analyse the images in real time. Don’t think you’ll need your powerful gaming rig or a server to get it to work - a Raspberry Pi Model 3 is enough for some sophisticated recognition already.
Software is the key to enabling Smart Cams. Generally speaking; the more efficient the code runs, the lower the processing power needed the run it.
Let’s look at what a Smart Cam can do
The analysis is where the Smart part comes in; with the processing power of modern technology, it is possible to do very complex instant recognition of anything passing in front of the camera;
- A soda bottle not completely sealed on the production line? The Smart Cam recognizes this, informs the plant manager so it can be corrected without production needing to stop. Plant manager not available? The production line is stopped before injuries occur.
- A builder enters the building site without a hard hat? A beacon flashes
- Two forklifts heading in each other’s direction? Buzzer sounds or forklift brakes get automagically applied
- Lights left on in the office after dark? System turns them off
- Office car park full? You get the gist
Of course, many of the applications above can be handled by using other technology like sensors and switches as well. What makes Smart Cams such a good alternative is the fact that they so generic. They can be used for any, no, ALL, of these applications and more. They can be repurposed, reprogrammed and do multiple tasks at the same time. Infrared camera’s open a whole new set of possibilities still.
All about anomalies
Most Smart Cams work on the basis of what is ‘normal’. Instead of programming what all the variations of anomalies looks, they focus on what normal looks like and react to those. All anomalies are flagged as they stand out from the pattern they scan for.
Imagine a situation where a certain colour vest must be worn in a factory (e.g. hi-res yellow at a warehouse) for safety reasons. With the Smart Cam installed at the entrance of the restricted area, it will verify all people passing are wearing that hi-res yellow. These people adhere to the ‘normal’ and the system won’t react. If you should forget to wear your vest and attempt to enter the area, the Smart Cam will detect you are an anomaly for not wearing hi-res yellow and raise the alarm.
Where do we go from here?
What if you could make all your currently installed security camera’s smart? Think about all the possibilities that would provide to make your office, factory, warehouse more efficient, safer and more cost effective to operate. Imagine that if a Raspberry Pi is able to do the detection processing, what you could detect and process when you put more powerful hardware in place, like a server or even IBM Watson!
The main thing you need is the right software for your situation. Software for facial or any other type of recognition is available as open source (e.g. https://facedetection.com/software/) and from several commercial vendors. Google is your friend here. It’s also possible to buy camera’s that are purpose built for recognition like the consumer focused Nest Cam IQ and Netatmo Presence.
It is likely that your situation will need a bespoke solution until that time that Smart Cams for business and industrial use become more mainstream, but don’t be surprised if Smart Cams become a commodity in the near future. We’ll be looking into this technology in more detail so keep an eye out for more on Smart Cams soon.