The deep learning phenomenon continues to excite the IT world, with computing power now at the level where it can be properly used in practical applications.

Hikvision has been at the forefront of applying the technology in the surveillance industry and beyond. It has already released its first set of products that harness the power of artificial intelligence (AI).

The concept of deep learning takes inspiration from the way the human brain works. Our brain is a very complex deep learning model. Brain neural networks are comprised of billions of interconnected neurons; deep learning simulates this structure. These multi-layer networks can collect information and perform corresponding actions according to analysis of that information.The ability of deep learning to enhance the recognition of human beings makes it a great addition to the security arsenal

Reducing false alarms

In the past two years, the technology has excelled in speech recognition, computer vision, voice translation, and much more. It has even surpassed human capabilities in the areas of facial verification and image classification; hence, it has been highly regarded in the field of video surveillance for the security industry.

Its ability to enhance the recognition of human beings - distinguishing them from animals, for example - makes the technology a great addition to the security arsenal. This is especially relevant in a world where false alarms account for 94%-99% of all alarms, according to police and fire service statistics.

Deep learning for data processing

Deep learning is intrinsically different from other algorithms. The way it solves the insufficiencies of traditional algorithms is encompassed in the following aspects.

The algorithmic model for deep learning has a much deeper structure than the traditional algorithms. Sometimes, the number of layers can reach over a hundred, enabling it to process large amounts of data in complex classifications.

Deep learning is very similar to the human learning process and has a layer-by-layer feature-abstraction process. Each layer will have different 'weighting', and this weighting reflects on what was learnt about the images’ 'components'.The more features there are, the more accurate the recognition and classification will be

Feature classification and recognition

The higher the layer level, the more specific the components. Just like the human brain, an original signal in deep learning passes through layers of processing; next, it takes a partial understanding (shallow) to an overall abstraction (deep) where it can perceive the object.

Deep learning does not require manual intervention but relies on a computer to extract features by itself. This way, it is able to extract as many features from the target as possible, including abstract features that are difficult or impossible to describe.

The more features there are, the more accurate the recognition and classification will be. Some of the most direct benefits that deep learning algorithms can bring include achieving comparable or even better-than-human pattern recognition accuracy, strong anti-interference capabilities, and the ability to classify and recognise thousands of features.

Conventional surveillance challenges

Conventional surveillance systems mostly detect moving targets without further analysis. Even smart IP cameras can only map individual points on a shape one by one, making it difficult to calibrate some features (e.g. forehead or cheek), thus decreasing accuracy. 

For perimeter security, for example, other technologies can be (and are) used to provide more comprehensive security. But they all have their downsides. Infrared emission detectors can be ‘jumped over’ but are also prone to false alarms caused by animals. Electronic fences can be a safety hazard and are limited in certain areas. Some of these solutions can also be expensive and complicated to install.Frequent false alarms are always an issue for end-users, who need to spend time to investigate each one

Identifying false alarms

Objects such as animals, leaves, or even light can cause false alarms, so being able to identify the presence of a human shape really improves the accuracy of perimeter VCA functions.

Frequent false alarms are always an issue for end-users, who need to spend time to investigate each one, potentially delaying any necessary response and generally affecting efficiency. 

Imagine, for example, a scenario where it’s relatively quiet – a location at night where there are few cars and people around. Even here, there could be 50 false alarms in a night. Assuming it takes 2-3 minutes to check out a false alarm, and that just 3 out of the 50 warrants more attention – say 15 minutes each.

A guard either needs to check the system and look back at the alert, or someone needs to be dispatched to the location and look around, checking if anyone has indeed ‘entered without permission’. In most organisations, these would need to be reported/recorded too, adding to the overall time spent on this ‘false alarm’.

Increasing recognition accuracy

Deep learning, however, makes a big difference. With a large amount of good quality data from the cameras and other sources, like the Hikvision Research Institute, and over a hundred data cleaning team members to label the video images, sample data with millions of categories within the industry have been accumulated. With this large amount of quality training data, human, vehicle, and object pattern recognition models become more and more accurate for video surveillance use.

Based on a series of experiments, the recognition accuracy of solutions using the deep learning algorithm increased accuracy by 38% - applying this to the previous example, that’s a saving of nearly one hour each night. This makes deep learning technology a great advantage in a perimeter security solution, with much more accurate line crossing, intrusion, entrance and exit detection.The value of deep learning technology stretches further than traditional security

Accurate motion detection

The value of deep learning technology stretches further than traditional security. For example, tracking movement patterns of individuals can see if they are ‘loitering’ and a potential threat in the future.

A threshold could be set to five metres radius of movement, or ten seconds of staying in the same place. If the person passes either threshold, an alarm could be triggered. The solution tracks the individual and compares this behaviour to a database to see if it recognises a pattern.

Another application would be in a scenario where ‘falling down’ could be a threat, like an elderly care home. If a height threshold was set at 0.5m and duration time 10 seconds, for example, the solution would be able to see a person falling down (as they go below 0.5m) and might be in trouble (if they ‘stay down’ for longer than 10 seconds).

AI-powered deep learning

The solution uses the parameters set to compare with its database and raise an alarm. With features and benefits like these, it’s easy to see how many smart applications could be catered for by deep learning technology.Artificial intelligence has massive potential

To sum up 10,000-strong R&D Centre is pushing the boundaries of surveillance solutions and bringing even more benefits to them. Artificial intelligence has massive potential, and Hikvision is always exploring new ways to apply this exciting technology throughout the security industry and beyond.

Intelligent search

Hikvision deep learning solutions are available at three levels:

  • DeepinView camera can conduct target tracking, grading and capturing when an alarm is triggered.
  • ‘Traditional’ IP camera using a DeepinMind NVR will add the function of searching intelligently by picture, saving time on searching for targets compared with a regular NVR.
  • DeepinView camera and a DeepinMind NVR delivers a full power solution, with the camera sending the information to the NVR, which can then analyse it. This accelerates recording and false alarm filtering.
Download PDF version

In case you missed it

Enhance traditional security systems within your smart home
Enhance traditional security systems within your smart home

Market dynamics are changing the U.S. residential security market, creating new business models that better appeal to the approximately 70% of households without a security system. Smart home adjacencies have helped revitalise the traditional security industry, and alternative approaches to systems and monitoring for the security industry are emerging, including a new batch of DIY systems. Growth in the residential security market and its position as the channel for smart home solutions have attracted numerous new entrants. Telecoms, cable operators, and CE (consumer electronics) manufacturers are joining traditional security players as they compete to fulfill consumer demand for safety and security. Connected products also provide a layer of competition as consumers must decide whether having category devices such as doorbell video cameras, networked cameras, and other products suffice for their security. Increasingly competitive landscape Smart home services can provide additional revenue streams for the security industry For instance, IP cameras are a highly popular smart home device rooted in security, and Parks Associates estimates 7.7 million standalone and all-in-one networked/IP cameras will be sold in the U.S. in 2018, with $889M in revenues. Product owners may feel their security needs are fulfilled with this single purchase, as such dealers and service providers are under increasing pressure to communicate their value proposition to consumers. Categorically, each type of player is facing competition uniquely—national, regional, and local dealers all have a different strategy for overcoming the increasingly competitive landscape. Smart home services can provide additional revenue streams for the security industry. In Parks Associates’ 2017 survey of U.S. security dealers, 58% report that smart home service capabilities enable extra monthly revenue. Almost half of dealers also note they have to offer smart home devices and services in order to keep up with their competition. While white-label devices are acceptable in some instances, dealers need to integrate with hero products whenever possible when those exist for a category. For dealers who have added smart home devices and services are all potential benefits and good for business Improved customer engagement That 2017 survey also revealed 36% of security dealers that offer interactive services report security system sales with a networked camera and 16% report sales with a smart thermostat. For dealers who have added smart home devices and services, enhanced system utility, increased daily value, and improved customer engagement with the system are all potential benefits and good for business. Security has served as the most productive channel for smart home solutions, mainly because the products create natural extensions of a security system’s functions and benefits, but as smart home devices, subsystems, and controllers expand their functionality, availability, and DIY capabilities, many standalone devices constitute competition to classical security. Particularly viable substitute devices include IP cameras, smart door locks, smart garage doors, or a combination of these devices. Products that are self-installed offer both convenience and cost savings, and these drivers are significant among DIY consumers—among the 6% of broadband households that installed a security system themselves, 39% did it to save money. Enhance traditional security Self-installable smart home devices may resonate with a segment of the market who want security While many security dealers believe substitute offerings are a threat, some dealers do not find such devices an existential threat but instead view them as another path to consumer awareness. They argue that the difference between smart product substitutes and traditional security is that of a solution that provides knowledge versus a system that gives one the ability to act on that knowledge. A common theme among professional monitoring providers is that a homeowner who is aware of events happening in the home does not necessarily have a secure and protected household. For example, a Nest camera, a DIY product, notifies a consumer via smartphone about events in the home when it detects motion, but only when the notification is opened and identified will a consumer be able to act on the related event. Self-installable smart home devices may resonate with a segment of the market who want security but are unwilling to adopt professional monitoring; however, providers can leverage these devices to enhance traditional security features and communicate the value of professional monitoring. Smart home devices and features, while posing a threat to some security companies, are a potential way forward to increased market growth Increased market growth A key counterstrategy for security dealers and companies is to leverage their current, powerful role as the prime channel for smart home devices. Many security dealers now include smart home devices with their security systems to complement their offerings and increase system engagement. For example, as of Q4 2017, nearly 70% of U.S. broadband households that were very likely to purchase a security system in the next 12 months reported that they want a camera to be included as part of their security system purchase. In response, many security system providers now offer IP cameras as optional enhancements for their systems. Smart home devices and features, while posing a threat to some security companies, are a potential way forward to increased market growth. Security dealers have an opportunity to become more than a security provider but a smart home solutions provider rooted in safety. Provide status updates Comcast has entered both the professionally monitored security market and the market for smart home services The alternative is to position as a provider of basic security with low price as the key differentiator. Comcast has entered both the professionally monitored security market and the market for smart home services independent of security. It has discovered that monetising smart home value propositions through recurring revenue becomes increasingly challenging as the value extends further away from life safety. Since the security industry remains the main channel for smart home services, security dealers are in a unique position to leverage that strength. Value propositions must shift from the traditional arming and disarming of a system to peace-of-mind experiences that builds off the benefits of smart devices in the home to provide status updates (e.g., if the kids arrived home safely) and monitoring at will (e.g., checking home status at any time to see a pet or monitor a package delivery). These types of clear value propositions and compelling use cases, which resonate with consumer and motivate them to expand beyond standalone products, will help expand the home security market.

What is the value of "free" video management systems?
What is the value of "free" video management systems?

They say that every choice has a cost. It's a basic principle that, economically speaking, nothing is free. If it doesn't cost actual money, it may be expensive in terms of time, attention and/or effort. These are interesting observations to keep in mind as one peruses the various "free" video management system (VMS) offerings available on the market. Some are provided by camera companies to unify their products into a "system", even if it's a small one. Other free VMS offerings are entry-level versions offered by software companies with the intent of the customer upgrading later to a paid version. For more insights, we asked this week's Expert Panel Roundtable: What is the value of “free” video management systems (VMSs) and how can a customer decide whether “free” is the right price for them?

The ongoing challenge of IT and data risk management
The ongoing challenge of IT and data risk management

Managing IT and data risk is a challenging job. When we outsource our IT, applications and data processing to third-parties more and more every day, managing that risk becomes almost impossible. No longer are our data and systems contained within an infrastructure that we have full control over. We now give vendors our data, and allow them to conduct operations on our behalf.  The problem is, we don’t control their infrastructure, and we can never fully look under the hood to understand and vet their ability to protect our data and operations. We have to fully understand how important this issue is, and ensure we have the right governance, processes and teams to identify and mitigate any risks found in our vendors. No longer are our data and systems contained within an infrastructure that we have full control over Today, everything is connected. Our own networks have Internet of Things (IoT) devices.  We have VPN connections coming in, and we aren’t always sure who is on the other end of that connection. It is a full-time job just to get a handle on our own risk. How much harder, and how much larger should our teams and budgets be, to truly know and trust that our vendors can secure those devices and external connections?  For every device and application we have internally, it is very difficult to even keep an accurate inventory. Do all of our vendors have some special sauce that allows them to overcome the traditional challenges of securing internal and vendor-connected networks? They are doing the same thing we are – doing our best with the limited human and financial resources allocated by our organisation. Risk stratification and control objectives  The benefits of outsourcing operations or using a vendor web application are clear. So how can we properly vet those vendors from an IT risk perspective?  The very first thing we need to put in place is Risk Stratification. Risk Stratification presents a few targeted questions in the purchasing process. These questions include – what type of data will be shared? How much of this data? Will the data be hosted by a vendor? Will this hosting be in the US or offshored? Has the vendor ever had a data breach? These questions allow you to quickly discern if a risk assessment is needed and if so, what depth and breadth.  Risk stratification allows you to make decisions that not only improve your team’s efficiency, but also ensure that you are not being a roadblock to the business Risk stratification allows you to make decisions that not only improve your team’s efficiency, but also ensure that you are not being a roadblock to the business. With risk stratification, you can justify the extra time needed to properly assess a vendor’s security.  And in the assessment of a vendor’s security, we have to consider what control objectives we will use. Control objectives are access controls, policies, encryption, etc. In healthcare, we often use the HITRUST set of control objectives. In assessing against those control objectives, we usually use a spreadsheet.  Today, there are many vendors who will sell us more automated ways to get that risk assessment completed, without passing spreadsheets back and forth. These solutions are great if you can get the additional budget approved.  Multi-factor authentication  Even if we are using old-fashioned spreadsheets, we can ensure that the questions asked of the vendor include a data flow and network/security architecture document.  We want to see the SOC2 report if they are hosting their solution in Amazon, etc. If they are hosting it within their own datacentre, we absolutely want to see a SOC2 Type II report. If they haven’t done that due diligence, should that be a risk for you?  Today, we really need to be requiring our vendors to have multi-factor authentication on both their Internet-facing access, as well as their privileged internal access to our sensitive data. I rate those vendors who do not have this control in place as a high risk. We’ve recently seen breaches that were able to happen because the company did not require administrators or DBAs to use a 2-factor authentication into sensitive customer data sources.  In the assessment of a vendor’s security, one has to consider what control objectives to use This situation brings up the issue of risk acceptance. Who in your organisation can accept a high risk? Are you simply doing qualitative risk assessment – high, medium and low risks? Or are you doing true quantitative risk analysis? The latter involves actually quantifying those risks in terms of likelihood and impact of a risk manifesting, and the dollar amount that could impact your organisation.   So is it a million dollars of risk? Who can accept that level of risk? Just the CEO? These are questions we need to entertain in our risk management programs, and socialised within your organisation.  This issue is so important – once we institute risk acceptance, our organisation suddenly starts caring about the vendors and applications we’re looking to engage.  If they are asked to accept a risk without some sort of mitigation, they suddenly care and think about that when they are vetting future outsourced solutions. Quantitative risk analysis involves quantifying risks in terms of likelihood and impact of a risk manifesting Risk management process  In this discussion, it is important to understand how we think of, and present, the gaps we identify in our risk management processes. A gap is not a risk. If I leave my front door unlocked, is that a control gap or a risk? It is a gap – an unlocked door. What is the risk?  The risk is the loss of property due to a burglary or the loss of life due to a violent criminal who got in because the door was unlocked. When we present risks, we can’t say the vendor doesn’t encrypt data. The risk of the lack of encryption is fines, loss of reputation, etc. due to the breach of data. A gap is not a risk.  Once we’ve conducted our risk analysis, we must then ensure that our contracts protect our organisation? If we’re in healthcare, we must determine if the vendor is, in fact, a true HIPAA Business Associate, and if so we get a Business Associate Agreement (BAA) in place. I also require my organisation to attach an IT Security Amendment to these contracts. The IT Security Amendment spells out those control objectives, and requires each vendor to sign off on those critical controls. We are responsible for protecting our organisation’s IT and data infrastructure – today that often means assessing a 3rd-party’s security controls One final note on risk assessments – we need to tier our vendors. We tier them in different ways – in healthcare a Tier 1 vendor is a vendor who will have our patient information on the Internet. Tiering allows us to subject our vendors to re-assessment. A tier 1 vendor should be re-assessed annually, and may require an actual onsite assessment vs. a desk audit. A tier 2 vendor is re-assessed every 2 years, etc. We are responsible for protecting our organisation’s IT and data infrastructure – today that often means assessing a 3rd-party’s security controls. We must be able to fully assess our vendors while not getting in the way of the business, which needs to ensure proper operations, financial productivity and customer satisfaction. If we truly understand our challenge of vendor risk management, we can tailor our operations to assess at the level needed, identify and report on risks, and follow-up on any risks that needed mitigated.