← Back to Articles

Top 8 AI features in CCTV solutions straight from the future

Ever since invention, CCTV solution have always felt like a futuristic thing. That fame was not quite without ground as it has really been ahead of its time from the technology perspective. However, there were some fantasies too mixed in which had added to that alien feel of video surveillance. Utopian science fiction novels telling us about the ever-present Big Brother watching everyone and everywhere, movies showing investigators that use camera footage to look around the corner and so on have added to the perception of CCTV solutions as something next-gen, cosmic. Cosmically compelling, cosmically complicated, and sometimes cosmically costly.

Just when the futuristic aura around video surveillance began to fade, artificial intelligence possibilities entered the scene, once again leaving everyone in awe. The rise of artificial intelligence popularity has boomed and bloomed, and has been thriving ever since. What we have now is only the beginning – and yet what we have now is already hard to believe is possible.

Here are some out-of-this-world features that are already available now, but make you feel they must have been sent from the future:

Remember the old joke: ‘Ever since I had learnt to walk, I’ve been always trying to lie down’. Much like that, ever since labor had made a human out of a monkey, the dream of creating a helper to delegate some hard or unpleasant tasks to has bothered our species’ mind. We’ve threaded a long but steady path to what we have now: computers can do computations at the speed of light, automated mechanisms can lift loads unimaginable to humans, robots can tirelessly do routine work without burning out. We’ve taught mechanisms to do repeated actions, but what they have been unceasingly striving (and failing) to have is the humans’ intelligence. Thinking like a person, vivid brain analysis, feelings, apprehension and reading one’s mind are yet to be learnt by machines, but there are some things that they can already do.

Analyzing people’s behavior for signs of mood seems easy to us, humans. If someone is smiling they are most probably having fun; eyebrows up means a person is amazed; and tears in someone’s eyes means they’re sad. Easy, isn’t it? But what if there are lots of people in an area, and you only have a glimpse before they pass by to check if there are signs of aggression or ill intents on their faces? What if there is a long distance between you and the people, so long that you can hardly make out their faces let alone their emotions? That’s what security guards in places of public gathering – like stations or airports – are supposed to be doing during long hours of their shift. Naturally, they could use a little help. And help do they get – with various CCTV solutions.

First, there was only a possibility to playback some suspicious moments: freeze time and have more moments to look an event over.
Then came zoom opportunities allowing to see objects in more details.
Now, advanced CCTV solutions can do live 24/7 analysis of the scenery with or instead of security officers, and trigger an alarm when disturbing signs in someone’s behavior are detected. For example, Xeoma intelligent video surveillance software can distinguish 6 basic human emotions: happiness, surprise, anger, disgust, fear, sadness, coupled with the neutral face expression.

The technology is now being actively utilized not just in security area: retail corporations pay loads of money to get to know their target audience, and behavior analysis, coupled with demographics information from other AI modules, tirelessly provide unbiased information, which is a level up compared to unreliable information provided by hired market analysis companies.

‘The eyes are the mirror of the soul’, says the well-known quote. As if inspired by this saying, neural networks technologies have directed their attention to looking into human’s eyes to better understand them.

Previously, eye tracking was solely prerogative of products aimed at helping people who lost their ability to move. The eye tracking products’ job was then to analyze movement of a pupil and interpret it into actions (for example, pressing a button on a website or scrolling a web page) or digits (for example, typing a message or search query).

Nowadays, the eye tracking field of application has expanded into the behavior analytics area. To some extent, the direction of a person’s gaze can help shed a light on what they think. Marketing departments use this technology to evaluate ads performance or to launch self-service kiosks or specific ads for those who look interested in a ‘bait’. With the information on what draws attention the most, merchandisers can select the best placement for high priority goods. The beauty of eye tracking is that the reaction is instantaneous, people often cannot suppress it in time even if they want so this information is probably the most true, unbiased, unlike other marketing queries conducted. After all, if a picture is worth a thousand words, one unconscious gaze is worth a thousand pictures.

If you are a fan of Sci-fi series, you will remember that these shows often show space ships with augmented reality rooms used for nostalgic recreation or business purposes like negotiations with far-away planet authorities. The only AR available in reality is leisure activities like 3D video games, most people would think. But that’s not entirely true. Driven by professional needs of security operators who frequently have to watch tons of footage to get the idea of how events unfolded, CCTV solutions have come to offer a similar technology for fast and quick yet efficient footage viewing, the time lapse.

Many VMS have the time lapse option in various kinds, but the idea behind this fast-forward playback is to track people and objects move around the scene. Such effect can be successfully fused with heatmapping (indication of most visited areas) or ways to highlight objects in the frame (for example, frames, arrows and dots, like in Xeoma, or multi-colored lines in some object tracking tools). Some VMS can make a kind of video summary, a collage of objects detected at different times in the same frame.

All in all, thanks to modifications of this technology, one glance is enough to see if a sought-for object ever appeared in camera sight which is super time-saving for search for missing/stolen items!

When we think of the police force of the future, we often imagine driverless patrol cars or drones cruising around and scanning the neighborhood for unusual behavior, a potential source of danger, or wanted criminals.

This vision is close to reality: even now various means for passive detection are used in police officers’ garments and patrol cars. For example, in-vehicle compact camera systems can do the real-time work of detection and recognition of license plates and generate alerts if a missing vehicle license flashes through road traffic. But that’s not all. Modern VMS can also ‘read’ signs and letters in the surroundings, recognize words, and alert the authorized personnel. This could be used for effortless background detection of graffiti or ads with inappropriate content.

This tool can also be used in business. Take retail for example: in a matter of seconds supervisors can use a smart device to scan shelves and get reports on whether the actual assortment is in compliance with management directions, or if the goods are mixed, which affects sales rate negatively. The feature can be combined with similar merchandising AI tools like the ones that alert personnel on some of shelves requiring replenishment of goods.

Sometimes smart text recognition tool is used for unfair competition where a competitor’s store is scanned to get quick and full information on their assortment and prices. However, mainly the tool is doing good, and shows a great potential to grow into self-checkout stores with no cashiers.

The previous section mentioned drones, and while we’re on the subject, have you noticed how many UAVs are flying over your house lately? Any public event like fireworks championships are now accompanied by drones filming. With so much information available, any person with engineer inclination can assemble and learn to fly a drone, and it is to be anticipated that there will be more of those who do – and consequently more of those who violate private property – knowingly or unknowingly. If you think this is not going to get on your nerves, just wait a couple of years when this phenomenon will become a massive scale.

Drone wars nowadays seem like a governmental field of expertise where terrorists use drones to bring threat, and state authorities invent counter measures to prevent and neutralise them. But drones might soon become too ubiquitous and shift from ‘their’ problem to ‘our’ type of problem. In other words, we are a step away from the future where correctional institutions will be on the lookout for drones trying to smuggle a ‘load’ into a prison, airport security systems will be needing to detect drones to avoid flight sabotages, and farmers will be forced to fight drones that scare cattle or fish in their reservoirs.

Drone detection in CCTV solutions relies on camera image rather than inception of a radio signal but luckily cameras evolution matches with the IT technology pace, so it’s not hard to find a camera with resolution good enough to cover even the vastest territory and see a drone from a distance.

Using bio metrics – be it finger prints, eye or face scan – to open doors, sign in computers or other automated mechanisms, is what we see in fiction movies a lot. While the Face ID feature that helps unlock a phone is quite commonplace these days, face identification in CCTV solutions is somewhat different. The double authentication technique that Face ID is a part of relies on advanced algorithms allowing to compare the face of a person in the camera sight with the ID card they are using for access. Two steps ahead of what we are used to, the two-step verification of the FaceID feature in video surveillance not only provides automated recognition of human faces, ridding us of the need to bother with physical keys, but also counters possible fraud attempts by requiring a QR code with the verified user profile that complies with the face in the frame.

The technology is especially popular in a university environment where access kiosks successfully replace human guards in a campus. This was especially valuable during trying times of a virus outbreak when social distance and limiting social contacts was essential to ‘flatten the curve’. When properly sanitized from time to time, such machines pose no risk of ‘catching’ and spreading a virus, thus saving lives of residents and guards – who otherwise would be at risk considering the number of contacts they undergo each day. Also, it leaves room for growth to be incorporated into access systems to top security areas like money vaults in a bank, or drug storage in a hospital.

The pandemic has changed our lives. Mask Detection is just one of examples of how well CCTV solutions reflect and respond to current events, needs and requirements of businesses and governmental entities, and how fast.
Anti-coronavirus measures like Mask Detection took less than half a year to be developed and employed (faster than vaccine!) and they might still be used further after the pandemic tumult lies down, for example, as a reinforcement in Safety Gear Detector package for chemical or pharmaseutical industries.

Although at first sight it might seem that the Mask Detection is as easy as it sounds, in reality it is internally quite an advanced tool accounting for different angles of head tilt and all the forms and color masks come in. This is one of the situations where machines do their best to ‘think’ i.e. analyse not only objects that look like the parent patterns but also extend detection onto something beyond that.
Anti-pandemics measures powered by artificial intelligence are a real-world example of how fast AI solutions can be erected in response to reality to help people adapt, minimize risks and even save lives. This might sound in a post-apocalyptic manner, however, change of reality might be a positive one too, which we witness every few decades, with breakthroughs like airplane flights, Internet, smart phones, etc. Come what may, humans will teach artificial intelligence to help.

Video surveillance solutions are mainly associated with ‘seeing’ – probably because historically it has been about getting camera footage and using it for post-incident investigation. However, an important aspect of many video security installations is capture of sound. Audio tracks can be used for service quality checks in call centers, or as a piece of evidence against misinformation lawsuits. Two-way audio opportunities available in some IP cameras and CCTV solutions can be utilized to send voice commands to the camera side – for example, a warning to trespassers or guidance to people stuck in an elevator, visitors at the door.

Even more than that, neural networks have made it possible to recognize various sound types, which can be a great standalone solution for long-distance sites, or a complimentary double measure for intrusion detection. For example, Xeoma’s Sounds Events Detector can distinguish between glass break, alarm, baby cry, screams and gunshots. Thanks to this set of sounds the tool has a great potential for security protection either in homes (intrusion detection), businesses (patient monitoring in hospitals), or on a city-wide level (for example, in Safe City programs).

These are just a few examples of how unearthly some of the current advancement feel. These technologies make us anticipate that the future of flying cars, hoverboards, plasma guns and Mars colonization is just around the corner. Pushed to the border of what’s deemed possible by ever-growing needs of more comfort, more optimization, more achievements, AI tools evolve fast. Technology of tomorrow are already around us and you can rest assured that more is coming!

13 July, 2022

Read also:
Xeoma: what it is and how it works
Additional modules in Xeoma
How AI in video surveillance helps fight Coronavirus