Category: Feature

With the growing number of schools in the United States with video surveillance, electronic access control and other traditional security systems, the number of potential “greenfield” projects for suppliers falls. Declining numbers of new security system installations will lead to slowing market growth over the next five years. Even so, security manufacturers are looking to implement the following new security technologies to improve school safety, which would also help reignite market growth: 1.     Facial recognition technology Many of the video surveillance systems currently used in schools are not actively monitored and also rarely lack any form of effective automated response. Adoption of facial recognition technology would allow the surveillance system to proactively search for potential threats, and alert school administrators and security staff about unrecognized individuals in the building. Unfortunately, affordable facial recognition technologies are often unable to adequately recognize the number of faces in a typical school; plus, these systems can place a large strain on a school’s information technology network. 2.     Logical and physical security identity management integration I ntegrating the school’s access control database with a higher authority logical database – for example, a student directory – would allow the access rights of former staff, and students who have been expelled or already graduated, to be removed automatically. However, access control providers may find it difficult to get permission to access student records and other sensitive data. Education administrators may also be uncomfortable with possibly creating a potential avenue of cyberattack. 3.     High-security classroom doors with multipoint looking systems Higher-grade doors would create a far more effective barrier between students and potential attackers, creating numerous safe spaces throughout the building in emergencies. It’s also true, though, that purchasing thousands of doors would be expensive. Fire regulations often dictate that key entrances and exits remain fail-safe during emergencies. Locking and unlocking doors multiple times would also disrupt teaching. 4.     Weapons checks using metal detectors or x-ray machines at school entrances Using metal detectors or x-ray machines at entrances along the school perimeter makes bringing weapons into the school much more difficult. However, schools often have multiple entrances, which means each school would require multiple detectors or x-ray machines – both of which are expensive. The school would also need to hire additional security staff to operate each machine. Securing entrances in this manner would also mean long queues would form after breaks and lunchtime, reducing the free time of students and staff. Access control intelligence service Entering its fourth full year, the IHS Markit Access Control Intelligence Service provides primary analysis of the equipment market, plus thorough investigation of key technology trends affecting the industry. Current topics of focus include mobile access, logical and physical identity management integration, adoption of biometric technology and security system convergence.

Beguiling, baffling or both – that’s blockchain. Aiming to clarify the subject for the benefit of companies and other organizations, the National Institute of Standards and Technology (NIST) has released a straightforward introduction to blockchain, which underpins Bitcoin and other digital currencies. Virtual barrels of digital ink are flowing in the media nowadays about these cryptocurrencies and the underlying blockchain technology that enables them. Much of the attention stems either from the giddy heights of value attained lately by the most well-known of these currencies, Bitcoin, or from the novelty of blockchain itself, which has been described (link is external) as the most disruptive technology since the internet. Blockchain’s proponents believe it lets individuals perform transactions safely without the costs or security risks that accompany the intermediaries that are required in conventional transactions. The NIST report’s authors hope it will be useful to businesses that want to make clear-eyed decisions about whether blockchain would be an asset to their products. “We want to help people understand how blockchains work so that they can appropriately and usefully apply them to technology problems,” said Dylan Yaga, a NIST Computer Scientist who is one of the report’s authors, “It’s an introduction to the things you should understand and think about if you want to use blockchain.” A blockchain is essentially a decentralized ledger that maintains transaction records on many computers simultaneously. Once a group, or block, of records is entered into the ledger, the block’s information is connected mathematically to other blocks, forming a chain of records. Because of this mathematical relationship, the information in a particular block cannot be altered without changing all subsequent blocks in the chain and creating a discrepancy that other record-keepers in the network would immediately notice. In this way, blockchain technology produces a dependable ledger without requiring record-keepers to know or trust one another, which eliminates the dangers that come with data being kept in a central location by a single owner. The blockchain idea has attracted enough supporters that there are now several hundred digital currencies on the market, and the companies that are investigating ways to employ blockchain number many more. Because the market is growing so rapidly, several stakeholders, customers and agencies asked NIST to create a straightforward description of blockchain so that newcomers to the marketplace could enter with the same knowledge about the technology. “Blockchain is a powerful new paradigm for business,” Yaga said, “People should use it, if it’s appropriate.” The question is when it is appropriate. As with any new tool, there can be a temptation to employ it purely for its novelty value. The report outlines some possible use cases, including banking, supply chain management and keeping track of insurance transactions. The report, Yaga said, was created partly to help IT managers make informed decisions about whether blockchain is the right tool for a given task. “In the corporate world, there’s always a push to adopt new technologies,” Yaga said, “Blockchain is today’s shiny new toy, and there’s a big push to adopt it because of that.” “We want to help people to see past the hype,” he said, “as lofty a goal as that is.” NIST has been tasked before with writing definitions of emerging concepts in information technology such as the definition of cloud computing it released in 2011. While Yaga describes the blockchain description as approachable – it’s “as high-level as I can write it,” he said – the document is longer than some other NIST definitions because the technology combines so many complex ideas. Among them are digital signatures, peer-to-peer networking and hash chains, all of which are tools common in cryptography and with which NIST has had extensive involvement. “We don’t have any axe to grind or product to sell, though,” Yaga said, “A lot of articles you’ll read online feature a disclaimer indicating that the author owns a certain amount of cryptocurrency or stock in a company. I have no vested interest in the monetary value of these networks. But we don’t pass judgment on the technology; we just want to get past the rumors.” To that end, Yaga said, the document began as a sort of FAQ addressing falsehoods the authors had come across such as the idea that there was no need for trust in the system. (“You do need trust,” he said, “just not a trusted third party, like a bank.”) It expanded to discuss the technical tools common to most blockchain-based systems and also explored related issues such as the high demands blockchain systems place on network resources. The roughly 60-page report might enlighten anyone who wants a picture of blockchain that is not skewed to any players’ interests, but Yaga said he and his co-authors hope it will give perspective to technical decision makers in particular. “A company’s IT managers need to be able to say, we understand this, and then be able to argue whether or not the company needs to use it based on that clear understanding,” he said, “Some people are saying you should use it everywhere for everything. We wrote with the perspective that you shouldn’t use it if it’s not necessary.”

Demand for professional video surveillance cameras has been growing quickly and is forecast to continue growing in 2018. It is estimated that less than 10 million surveillance cameras were shipped globally in 2006, which grew to over 100 million in 2016, and is forecast to make over 130 million during 2018. Despite this increase in demand, the average price of cameras and other video surveillance equipment will continue to fall quickly. As a result, IHS Markit forecasts that in terms of US dollar revenues the world market for video surveillance equipment will grow at an annual rate of less than 6% in 2018. It will be challenging for vendors to continue to grow revenues and margins, but there will be opportunities for well-placed vendors. For example, both the South East Asian and Indian markets are forecast to grow at higher than average rates. There is also great potential for the next generation of products powered by technologies like deep learning and cloud computing. So, what will be the big stories during 2018? Deep learning, GDPR compliance and drone detection technologies are just some of the trends discussed in this eighth annual trends IHS white paper. The following articles are designed to provide some guidance on the top trends for 2018 in the video surveillance industry.     By Jon Cropley   The A to I of Video Surveillance Terminology The past 12 months have seen a range of new terms becoming regularly used in the video surveillance industry. We attempt to provide a brief summary of some of these. AI (artificial intelligence): Computers are able to perform specific tasks as well as, or even better than human intelligence. In the context of video surveillance, AI is used in the field of computer vision to classify visual images and patterns within them. Big data: Huge amounts of different information are  stored, organized and analyzed by computers to identify trends, patterns, and relationships. In the context of video surveillance, the data could be metadata describing hours of video surveillance footage combined with other data sources to highlight patterns relating to security or business operations. Cloud computing: Instead of using a local server to store or manage video surveillance data, use a network of internet-connected remote servers. Generally this network has the ability to provide additional resource if and when required from a larger available pool. The available resource may be clustered into a datacenter or network of datacenters. These may be private (entirely or partly owned for exclusive use by specific organization/s) or public (resource accessible to multiple separate users). Deep learning: A branch of machine learning and subset in the field of AI. Deep learning makes use of algorithms to structure high-level abstractions in data by processing multiple layers of information, emulating the workings of a human brain (a neural network). Edge computing/ storage: Performing data processing and analytics/ storage closest to the source of the data (normally, in this context, in a video surveillance camera). Face recognition: When a video surveillance system can automatically match a person’s face against a database of individuals. GPU (graphics processing unit): A programmable chip specialized for use in image processing. Due to the requirement to be able to simultaneously processing multiple large data blocks required in modern image processing, GPUs have been found to be highly suitable for deep learning/ neural network processing. H.265 (or MPEG-4 part 2): H.265 is a video compression codec standard approved by the International Telecommunications Union (ITU-T). Compared with H.264, H.265 has the potential to use 30-40% less bandwidth for a video stream of the same quality. IoT (Internet of things): IoT is not a specific device or technology – it is a conceptual framework, driven by the idea of embedding connectivity and intelligence in a wide range of devices. IHS Markit defines an IoT device as a device which has some form of embedded connectivity that allows the device to be directly connected to the internet (i.e., IP addressable), or allows the device to connect (tether) to an IP addressable device. In the context of video surveillance, this could be using video surveillance data with other sensors or sources of information.   Big Differences between the Chinese Market and the Rest of the World China is forecast to account for over 46% of global professional video surveillance equipment revenues in 2018. Despite this, the Chinese market has some unique characteristics that make it very different to other regional markets. This has led to the suggestion that there are two markets for video surveillance equipment – the Chinese market and the world market excluding China. Below are 5 ways in which the Chinese market differs from the rest of the world. Supply of equipment is more concentrated in China The two largest vendors of branded video surveillance equipment accounted for over 50% of the Chinese market in 2016. This compares to the world excluding China where the two largest vendors accounted for less than 20% of the market. Shipments of deep learning-enabled equipment are much higher in China It is forecast that three quarters of all deep learning-enabled servers for video surveillance shipped worldwide in 2018 will be shipped in China. Domestic vendors dominate supply of equipment in China Chinese vendors account for more than 80% of Chinese market revenues. There are other countries where domestic brands dominate supply (Germany, Japan and South Korea are all examples). However, in most other country markets, foreign vendors account for a much higher proportion of revenues. Shipments of HD CCTV are proportionally lower in China It is forecast that HD CCTV cameras will account for only around 10% of all cameras shipped in China in 2018. It is forecast they will account for over half of all camera shipments in the world excluding China. The market has been growing faster The Chinese market has grown at an average annual rate of 13.3% between 2012 and 2017. In comparison, the World market excluding China has grown at an average annual rate of…

2017 will undoubtedly be considered as the ‘Year of ransomware,’ the year when the global security landscape was forever changed by attacks like WannaCry and NotPetya. The havoc caused by these attacks reached far beyond the paltry ransom demand. Hospitals turned away patients, production lines came to a halt, nuclear radiation monitoring was disrupted, and so on. Cyber events like these were a wake-up call to the brave new world of cyberattacks and how they could reach further into the ‘real world’ than ever before. But it’s not just ransomware and the threat landscape that have changed; digital transformation initiatives like the move to the cloud and the increasing convergence of IT and operational technology (OT) have drastically changed networks and expanded responsibilities of security teams tasked with protecting them. These initiatives, for all their business benefits, have caused network complexity to skyrocket – an issue cyberattackers are all too eager to exploit. All the while, the labour force to safeguard against attacks remains stubbornly inadequate. As we head into 2018, here’s a look at the cybersecurity trends sure to emerge amid the intersection of the most capable threat landscape, the most complex networks and a worldwide – including India – skills shortage. Hybrid networks stretch attack surfaces The attack surface – the total sum of the ways an organization is susceptible to cyberattack – is like a balloon. It expands with the introduction of new attack vectors and attack targets, like extensions into virtual, cloud and operational technology environments. It contracts with the good cyber hygiene and risk reduction. The larger your attack surface, the more likely it is to pop. To control an attack surface that encompasses hybrid networks and that’s affected by a constantly evolving threat landscape, organizations will need to unify visibility and centralize management. Gaining seamless visibility across physical IT and OT networks, as well as virtual and cloud networks, will give them a holistic foundation on which to build a security program ready to address an agile threat landscape. While different technologies, processes and teams may be involved to secure various types of networks, attackers don’t pay attention to such divisions. If anything, they exploit these divisions in security management, and simply follow the path of least resistance to reach their intended target, no matter where the attack originates. In 2018, we’ll likely see attackers leverage hybrid network connectivity to infiltrate cloud and OT networks where traditional cybersecurity measures are still being fleshed out. Ongoing evolution of distributed attacks While NotPetya was originally dubbed a ransomware attack, seemingly a new iteration of WannaCry, it was pretty lousy at the ransom part. What it and WannaCry demonstrated, though, was the distributed attack model on which modern ransomware relies. It targets as many victims possible, looking for low – hanging fruit, so attacks can be carried out easily and automatically, maximizing the attacker’s ROI. Ransomware is a perfect fit for this model, in that any target can be extorted for payment. Now that the distributed attacks have proven global-reach capabilities, we’re sure to see more mass-scale attacks in 2018. The good news about the distributed attack model is that the cyber-criminals sell or share proven attack methods such as vulnerability exploits, to carry out distributed attacks. Again, this practice gives them a better ROI than developing new, native exploits or exploits for a specific target. It also means a relatively small number of vulnerabilities exploits are being used and reused. If vulnerability management programs can take a threat – centric approach and focus on this subset of vulnerabilities, they’ll have a greater impact on their organization’s security than if they targeted only CVSS critical vulnerabilities. Security goes automated out of necessity As mentioned earlier, networks are growing increasingly complex, meaning that IT security teams must contend with growing amounts of data that needs to be contextualized, analyzed and acted upon. At the same time, the industry is suffering from a worrying talent shortage, which means that there are fewer skilled workers available to manage these issues. This is creating an environment where attack vectors abound, increasing the organization’s risk of attack. As a result, in 2018 we expect to see a surge in the adoption of automated solutions, particularly for integrated analytical workflows. These can deliver actionable intelligence to security practitioners of what to focus on – such as vulnerabilities posing an imminent threat – what tools are at their disposal to take action, and tracking the workflow to ensure tasks are carried out to completion. The scales tip to the Cloud At present, most firms are in a transition phase, with networks made up of a hybrid of physical, virtual and multi-cloud environments. 2018 is set to be the tipping point, as corporate networks become predominantly or entirely virtual or cloud-based. Organizations will need to be sure they understand and can support the shared responsibility model of the cloud, in that the cloud service provider is responsible for security of the cloud while the organization is responsible for security in the cloud. The assumption that cloud networks are inherently secure needs to be overcome, and security teams need to have the means to understand how traffic moves into, out of and within cloud networks to put the proper security controls in place. The importance of automation, as mentioned above, becomes even more important in cloud-networks. The elasticity of clouds makes their security management too much to handle via manual processes. If organizations are aiming for a complete move to the cloud, they need to ensure that security programs are poised to support the approach. Increased attacks on operational technology The convergence of IT and OT networks presents several advantages in terms of productivity, ease of management and cost-effectiveness. But it has also introduced new cyber risks to critical infrastructure organizations such as utilities, energy producers and manufacturers that could have very real impacts on uptime, human safety and the environment. Because of the havoc that can be caused, OT networks have become an attractive…

In this ever-expanding era of artificial intelligence (AI), Deep Learning will soon become the foundational technology for the security industry. Technologies that ‘learn’ will become more common and more powerful. This trend will strengthen critical security efforts in every sphere. Hikvision introduces three camera models equipped with deep learning algorithms in the smart retail industry. With the growing popularity of shopping online, the retail sector has felt the disruptive impact of Internet e-commerce more than most industries. Some have reacted to online competition by closing physical stores, but others are attempting to overcome challenges through technological transformation. Traditional retail lacks intelligent tools for accurate data collection and visualization, making it unable to provide a basis for business decision-making at the shop. People counting There is an old saying in the trade industry – ‘Small profits but quick turnover;’ and footfall is a ‘KPI’ – key profit indicator – that helps make that turnover. Compared to e-commerce, traditional offline retail stores lack the capabilities to accurately calculate customer flow. Hikvision’s dual-lens people-counting camera provides accurate customer counting and generates customer flow trends to evaluate performance and strategic initiatives. Although in a real-world scenario, shadows or other objects may easily cause miscounts, but the dual-lens people-counting camera, equipped with two cameras and powered by a deep learning algorithm, easily overcomes such interferences and delivers highly accurate data. A key advantage of deep learning algorithms over surveillance cameras’ vision algorithms is that deep learning can be continuously trained and improved with better and more data-sets. This means the longer it works, smarter it gets. Featuring binocular stereo vision, 3D people detection, and height filtering technologies, the dual-lens people-counting camera is able to accurately distinguish human beings from non-human objects, and movements in the background. By analyzing customer flow data, store management can optimise the allocation of the workforce to ensure better customer service and reach higher profits. Store managers can schedule staff strategically for peak and off-peak hours. Furthermore, they can also develop strategic marketing activities to attract customers by analyzing the data of incoming rates (entering vs. passing by). Heat mapping When customers enter the store, retailers are concerned to know about what merchandise customers are interested in. Before that, what’s more important is how to get the route they walk through and where they stop. With Hikvision’s Heat Mapping function, retailers can determine the amount of time shoppers spend in specific areas of a store, identify hot spots and dead zones, and measure the number of people who actually shop for specific products, rather than just casually walk by. Heat mapping is used to monitor and measure the size of target traffic in a region. It is a graphical representation of data represented by colors, and it is usually used to analyze the visit times and dwell times of customers in a specified area. The heat mapping function is often used in shopping malls, supermarkets, museums etc., that can find customers’ preferences over time through heat maps, offering insight how to best place items and design the store layout. As a representative product, Hikvision’s fisheye cameras, equipped with heat mapping function, not only capture a panoramic high-definition image but also learn about heat conditions in different regions within a store. In spacious areas, fewer cameras means reduced installation and labour fees. Hikvision’s  fisheye cameras are ideal for these areas, maximising monitoring views and image quality insurance. Queue detection What do you do if you always have to queue for a long time while shopping? Keep waiting, or give up shopping and avoid that store in the future. In the retail industry, waiting time is one of the most important factors affecting the customer experience. Hikvision’s Queue Detection function can help retailers manage checkout lines. When too many customers enter a queue, it can notify management to open a new checkout line. More specifically, Hikvision’ queue detection cameras can monitor the queuing situation in real-time. Firstly, cameras count the number of people in each queue and then track the dwell time of each customer. Once it is found that the number of people in a queue is too many, or the average dwell time of customers is too long, an alarm is triggered to prompt a response. Store management is reminded to open checkout windows to reduce waiting times, improving transaction efficiency and the entire shopping experience. Hikvision smart retail solution is designed to help retailers bring offline stores into a digital world, allowing data to support management and operations. It promotes retailers’ technological transformation in response to increased industry competition through the use of innovative retail technology. Ashish P. Dhakan – MD & CEO, Prama Hikvision    

Demand for professional video surveillance cameras has been growing quickly and is forecast to continue growing in 2018. It is estimated that less than 10 million surveillance cameras were shipped globally in 2006, which grew to over 100 million in 2016, and is forecast to make over 130 million during 2018. Despite this increase in demand, the average price of cameras and other video surveillance equipment will continue to fall quickly. As a result, IHS Markit forecasts that in terms of US dollar revenues the world market for video surveillance equipment will grow at an annual rate of less than 6% in 2018. It will be challenging for vendors to continue to grow revenues and margins, but there will be opportunities for well-placed vendors. For example, both the South East Asian and Indian markets are forecast to grow at higher than average rates. There is also great potential for the next generation of products powered by technologies like deep learning and cloud computing. So, what will be the big stories during 2018? Deep learning, GDPR compliance and drone detection technologies are just some of the trends discussed in this eighth annual trends IHS white paper. The following articles are designed to provide some guidance on the top trends for 2018 in the video surveillance industry.     By –   Oliver Philippou Drone Detection Technologies In last year’s ‘Video Surveillance Trends for 2017’ white paper, IHS Markit identified the use of drones and robots as a growing technology for security applications. In reality the use of drones (unmanned aerial vehicles) for this purpose has not really taken off because of three main issues – legislation, battery life and cost. However, the use of drones has emerged as a very real problem for perimeter security. For would-be intruders or unwanted observers using drones, the issues suffered by the commercial market do not apply. Consumer drones are readily available for just a couple of hundred dollars and flown by anyone with no prior training and without a license. Thus the problem of drones in restricted airspace such as near airports, critical infrastructure facilities, or above sports stadiums full of spectators, has become an increasing concern. Given the large physical area that these restricted airspaces cover, simply being able to identify that a drone is nearby has proven challenging. However, IHS Markit expects that recent developments in drone detection technology mean that in 2018 anyone wanting to secure a perimeter will have to take into account the threat from above. There are currently two main drone detection technologies: Radio Frequency (RF) detection aerials: These products are used to detect, analyze and locate the RF used by the drone base station to communicate with the drone. The benefits of RF detection aerials are that they can detect and locate both the drone in the air and the person controlling the drone, in some instances even before the drone takes off. RF aerials are also able to detect over a much greater range than radar or video surveillance cameras. Additionally, RF aerials can potentially even interfere with the communication between the drone and the controller. However doing this can have unknown effects, such as the drone dropping out of the sky. The problem with RF aerials is that they are not particularly accurate at providing a specific location, nor are they able to provide video verification of a drone. Radar: Ground surveillance radars are used for low-level aerial surveillance to detect and track small drones. They use a similar technology to air traffic control radar, but on a smaller scale, providing a shorter operating range than RF aerials. However, radar is significantly more accurate than an RF aerial, and therefore is well-suited to be paired with video surveillance cameras including PTZ cameras automatically to zoom in and follow targets. The downside of radar is that performance can be impacted by adverse weather (for example rain, snow, sand or dust). Radar-based drone detection systems tend to be more expensive than RF aerial systems. Video surveillance alone is not considered an adequate detection technology. Object detection analytics can be used on a video surveillance camera for video verification and capturing a visual image of the potential threat. However, both RF aerials and radar can detect beyond the line of sight, and have nearly a 360-degree field of detection. Therefore many vendors are finding that combining all three technologies together offers the best chance of drone detection. The use of video has an important role to play in this emerging field of perimeter security.   IHS Analyses  The A to I of Video Surveillance Terminology    By  – Jon Cropley Big Differences between the Chinese Market and the Rest of the World  By –  Jon Cropley General Data Protection Regulation (GDPR)   By – Josh Woodhouse Video Surveillance Fault Tolerance   By – Josh Woodhouse Forensic Video Analytics as a Service   By – Josh Woodhouse The Evolution of Deep Learning in Video Surveillance   By –  Monica Wang

Demand for professional video surveillance cameras has been growing quickly and is forecast to continue growing in 2018. It is estimated that less than 10 million surveillance cameras were shipped globally in 2006, which grew to over 100 million in 2016, and is forecast to make over 130 million during 2018. Despite this increase in demand, the average price of cameras and other video surveillance equipment will continue to fall quickly. As a result, IHS Markit forecasts that in terms of US dollar revenues the world market for video surveillance equipment will grow at an annual rate of less than 6% in 2018. It will be challenging for vendors to continue to grow revenues and margins, but there will be opportunities for well-placed vendors. For example, both the South East Asian and Indian markets are forecast to grow at higher than average rates. There is also great potential for the next generation of products powered by technologies like deep learning and cloud computing. So, what will be the big stories during 2018? Deep learning, GDPR compliance and drone detection technologies are just some of the trends discussed in this eighth annual trends IHS white paper. The following articles are designed to provide some guidance on the top trends for 2018 in the video surveillance industry.     By –  Monica Wang The Evolution of Deep Learning in Video Surveillance In last year’s edition of our annual trends white paper, ‘Top Video Surveillance Trends for 2017,’ it was discussed that the biggest challenge for mass adoption of deep learning was the ability to demonstrate a security or business intelligence benefit to using the technology in the many different surveillance scenarios. 2017 witnessed great progress in the market with a transition from proof of concept deep learning algorithms to video surveillance products and a whole range of new entrants in AI chipset offerings. With the technology’s concept more proven, future success will depend on the ability to demonstrate a return on investment from deployments. Driven by the R&D investment from chip vendors, software startups and major video surveillance vendors, deep learning video analytic algorithms have been developed into fully deployable products with user-friendly interfaces and scenario-focused solutions. For example, deep learning face recognition algorithms are now available in search engine type applications, designed to find missing people from video footage. Transitions such as these are evident in the Chinese market and in the products shown at the recent China Public Security Expo (CPSE) 2017. Full deep learning products on display were either software-based applications with deep learning or video surveillance hardware with embedded algorithms. As an increasing number of vendors develop deep learning algorithms, several software startups have also developed their own deep learning video surveillance hardware to cement their place in the market. Transformation in deep learning cameras Following the transition from analog to network cameras, the next stage will likely be a mass market transformation to deep learning enabled cameras. During the transition to network cameras, growth in shipments was accelerated due to large price declines. The worldwide average price of network cameras in 2016 was around one quarter of the 2010 level. A similar trend of large price decline catalyzing a rapid increase in unit shipments can also be expected for the future generations of deep learning enabled cameras. So far, most of the deep learning cameras sold have been for safe city projects run by police departments in China. These projects are less price sensitive than the remainder of the market, where the average price is still too high for end-users. The cost of semiconductors which enable the deep learning algorithms to run in the cameras are a major component of camera prices. Following the release of deep learning cameras with Nvidia and Movidius chip solutions, more semiconductor vendors (including some from the mobile device market) are highlighting their ambitions for the video surveillance market. Some of these vendors include XILINX, DeepHi Tech, Intel, Vimicro and Qualcomm. These new chip vendors entering this market are increasing the number of options available for the deep learning ecosystem and importantly are increasing pricing pressure at the chip level. This will enable a rapid reduction in the average price of deep learning cameras. Outlook Deep learning and AI are now more established buzz words, particularly in China, and the education of the market regarding the technology continues to increase. End-users are becoming more familiar with real world product deployments rather than just prototype demonstrations of an algorithm. Chinese vendors have begun to promote their deep learning products to the rest of the world. 2018 is set to continue this trend with increased sales from the Chinese vendors outside the Chinese market and more case studies from installations outside China. The year ahead will also see greater differentiation of video analytics products based on an increased number of semiconductor vendors’ chipsets. Besides the initial projects in city surveillance and transportation, more installations in retail and commercial buildings are likely to be the next to embrace the greater use of deep learning technology. As we’ve seen in the wider video surveillance market, a targeted vertical approach is likely to be a common strategy. Vendors that market vertically-focused deep learning applications aligned with their own existing portfolios should have good opportunities to grow.   IHS Analyses  The A to I of Video Surveillance Terminology    By  – Jon Cropley Big Differences between the Chinese Market and the Rest of the World  By –  Jon Cropley General Data Protection Regulation (GDPR)   By – Josh Woodhouse Video Surveillance Fault Tolerance   By – Josh Woodhouse Forensic Video Analytics as a Service   By – Josh Woodhouse Drone Detection Technologies   By –  Oliver Philippou

Demand for professional video surveillance cameras has been growing quickly and is forecast to continue growing in 2018. It is estimated that less than 10 million surveillance cameras were shipped globally in 2006, which grew to over 100 million in 2016, and is forecast to make over 130 million during 2018. Despite this increase in demand, the average price of cameras and other video surveillance equipment will continue to fall quickly. As a result, IHS Markit forecasts that in terms of US dollar revenues the world market for video surveillance equipment will grow at an annual rate of less than 6% in 2018. It will be challenging for vendors to continue to grow revenues and margins, but there will be opportunities for well-placed vendors. For example, both the South East Asian and Indian markets are forecast to grow at higher than average rates. There is also great potential for the next generation of products powered by technologies like deep learning and cloud computing. So, what will be the big stories during 2018? Deep learning, GDPR compliance and drone detection technologies are just some of the trends discussed in this eighth annual trends IHS white paper. The following articles are designed to provide some guidance on the top trends for 2018 in the video surveillance industry.     Josh Woodhouse Forensic Video Analytics as a Service When reviewing video for investigations one of the biggest challenges is the sheer volume of video footage which may need to be examined. It’s said that typically it may take a trained officer or analyst using traditional methods (a notepad and the pause/ rewind buttons) 1.5 – 2 hours to review just an hour of raw video footage. This can be a huge consumer of resource. The problem is particularly prominent in police forces where the issue is amplified by a combination of budgetary constraints and a spike in the amount of video inputs (increased use of body worn cameras and more publically submitted videos). There have long been grounds to find a more efficient solution. Several video surveillance analytic solutions for forensic video analysis have been available for some time, yet the improvement in accuracy in the last 18-24 months using deep learning technology has been critical. This advancement has pushed accuracy to a level of competency that can be reliable enough to assist human analysts. However, deploying this technology can prove expensive for police departments. At present there is a significant hardware cost required to run this type of video analytic. And many of the potential clients are not managing live cameras but instead looking for a tool to search through the repository of potential evidence they have collected from multiple sources in many formats. Some providers have offered use of their analytics and software packages in a ‘as a service model’ where police forces or agencies can utilize the vendor’s onsite infrastructure and internal analysts to outsource their forensic video analysis. Moving this model to a cloud platform is an obvious evolution where by following some training clients can use on-demand forensic video analytics for particular cases remotely with their own analysts without large hardware investment. This is an exciting prospect for some smaller forces, which may not have the available capital for their own infrastructure or a large enough case load to justify a large capital expenditure. For large agencies and police departments with either highly sensitive data and/ or large case-loads, their own onsite infrastructure will most likely be the most cost effective solution. IHS Markit expects that forensic video analytics will be integrated into existing cloud services. For example in the body-worn camera market many police forces already utilize the cloud to store and review body-worn video, yet, in these repositories we still see a degree of separation from other video sources, for example from fixed (public or private) video surveillance. In 2018 IHS Markit expects to see increased convergence in post recording video repositories where, even if only on case by case basis, multiple sources of video will be brought together to be investigated using deep learning video analytics for which cloud may be an important enabler for on demand requirements.   IHS Analyses  The A to I of Video Surveillance Terminology    By  – Jon Cropley Big Differences between the Chinese Market and the Rest of the World  By –  Jon Cropley General Data Protection Regulation (GDPR)   By – Josh Woodhouse Video Surveillance Fault Tolerance   By – Josh Woodhouse The Evolution of Deep Learning in Video Surveillance   By –  Monica Wang Drone Detection Technologies   By –  Oliver Philippou