Martin Gren is the co-founder, Axis Communications AB and the inventor of the world’s first network camera, which celebrated 15 years in 2011.
It’s very difficult to make predictions, especially about the future. Just look at Nostradamus—one of the most scrutinized figures in all of history. But at least with technology, there are rules and trends that you can rely upon as guides.
One of these trends is Moore’s law from 1965, which states that performances of computer-based hardware will double every two years. This is still valid today. In fact, since we at Axis Communications introduced the world’s first network camera in 1996, network camera performance for resolution and frame rate has increased by roughly 600 times in those 15 years. Moore’s law only suggests about 400 times improvement during that time period, so, in video surveillance, performance has overachieved!
But we can’t sit back and relax and expect Moore’s law to do all the work. While it keeps on working and will give us a lot of cool options for the future, there are areas of video surveillance—like the evolution of lenses and mechanics—that do not mirror Moore’s law. Here we have to look for true innovation instead.
Remember that we work in a slow-moving market of video surveillance with respect to technology. A good reference is the evolution of analog video, an invention from the 1940s. The latest and last real improvement to analog video was the addition of color in the 70s.
Thankfully the industry is tipping to all IP, where we’ll see many more (and much faster) innovations. Here is my list of predictions, beginning with technology driven ones.
Where’s my storage?
In the next 15 years, IP video will be available “to the people.” Currently, IP is the answer for larger systems, and typically is the only technology specified in systems larger than 32 cameras. The smaller systems, like the ones found in convenient stores, gas stations, boutiques, offices and franchise locations, still often rely on analog. However, the total cost of ownership (TCO) for IP in small systems continues to fall and four factors will accelerate this change in the coming years: large on-board storage, much better image processing, better compression and a higher speed Internet.
The latter is important as it makes Video-Surveillance-as-a-Service offerings (sometimes referred to as hosted video) possible and can nearly totally eliminate the need for onsite storage. In particular this is useful for chains of smaller stores, restaurants, etc. who can watch many dispersed and remote stores from one workstation or mobile device.
Increased capacity of on-board storage also offers ability for high-quality redundant storage and can even enable self-based recording systems where the network camera is the combined camera and NVR.
Analytics that actually work
We all remember the hype in video analytics and the promises of finding bags (but not carts) left empty at airports. We heard the same promises of distinguishing between the good guys and a terrorist and being able to pick them out in a crowded stadium. This promise was made despite the fact that a face in the distance is only made up of a few pixels. Despite the bad initial reputation, there are useful analytics working today if expectations are properly set.
The first useful analytics are functions such as high quality motion detection, camera tampering detection and cross line detection. These functions are very useful for security today and continue to improve.
There are other useful analytics that can be used for other parts of the business beyond security, such as marketing and building management. In these cases the analytics can afford a five percent failure rate without causing a disaster for the business or a security vulnerability. Analytics used in retail such as people counting, heat map generation and detection for empty shelves and product sweeps are expanding.
Now that we’ve passed the era of overhyped analytics, software developers can leverage the ever-growing processing power of the cameras (which, remember, are growing alongside Moore’s law) to deliver useful algorithms for different surveillance markets. It will be up to the integrator to learn when to sell the analytics as a feature.
Higher quality images and light sensitive cameras
Image sensors are evolving at a rapid pace and there is basically no more research and development in standard definition (SD) image sensors. The shift in the market to high-definition resolution and quality is now a fact and a major driver for IP.
For the coming year we will see analog cameras being taken down as end-users opt to move to 720p HDTV and then later to 1080p. There will be niche markets for super-high resolutions, but the industry needs a common standard and HDTV is a good plateau.
The human eye still beats the majority of all video cameras in many aspects, but the turning point where a network camera can see better than the human eye is being passed. Today there are technologies that beat the eye in various aspects—such as super-low light color cameras, improved dynamic range (WDR/HDR) and thermal/IR cameras. But still the human eye is superior in the combination of all these scenarios as well as in identifying objects. Now this is a challenge for all research and development people out there!
The second aspect of future innovation lies in new applications for surveillance cameras. In the same way as it was almost impossible to execute city surveillance with analog in the efficient way we do today with IP, we will see the technology give us new usages in other markets.
In retail we typically have the highest pace of innovation of alternate camera usage as retailers have so many cameras and different needs. If the return on investments can be proven, it scales up quickly. I think we will see a lot of innovation here based on analytics use but also on supply-chain management and merchandising. Imagine you are a brand owner and want to see how your brand is exposed in certain stores. This is possible to do if you have low-cost PTZ network cameras that can be shared remotely by the brand owners as well as by creating a closed retail system with limited partner access through a hosted video model. Applications such as dwell time analysis, shopping unit counting and tracking/handover to map customer behavior will be commonplace among major retailers.
Other areas for increased camera usage are in city surveillance and transportation. Airports are already full of cameras but they keep on multiplying. But in bus and rail transportation we see a lot of opportunities as these verticals are not as penetrated as they could be.
And of course, hosted video
Finally I see video-as-a-service becoming a monster trend. This is something I have been evangelizing for quite a while and today we see many of the pieces falling together. Even many of the largest integrators are pushing hosted video.
After all, we trust the cloud for our emails, documents and even finances. Why shouldn’t the cloud offer us video surveillance as well? This, in combination with local storage in cameras, will begin a revolution in the world of surveillance for smaller sites where analog still dominates. This will be the main driver towards reaching 100 percent penetration for IP cameras in the world, which could be as early as 2020 based on today’s pace for new installs.
Hosted video will bring us many new usages of video that we don’t know right now. Alarm verification with video, construction site monitoring and even city surveillance systems will all benefit from hosted solutions. And maybe even our homes will be getting more cameras in the coming 15 years.
Martin Gren is the co-founder, Axis Communications AB and the inventor of the world’s first network camera, pictured on page 26 with Gren, which celebrated 15 years in 2011.