I still say we should go back to VCRs. Impervious to hacking!
Video Surveillance History, 2000 - 2023
The video surveillance market has changed significantly since 2000, going from VCRs to now AI and the cloud.
The goal of this history is to help professionals understand the important business and technology shifts that impact the market today, including:
- 2000 - 2005 DVR Era
- 2001 - 9/11 Impact
- 2006 - Infancy IP and VMS
- 2008 - 2012 MP Cameras Go H.264
- 2009 - 2013 Cloud Hype / Bursts
- 2010 - 2018 Struggles For Video Analytics
- 2012 - 2014 Rise and Fall of Edge Storage
- 2010s WDR and Low Light Improvements
- 2015 Smart CODECs Rise
- 2018 H.265 Mainstream
- Storage No Longer Major Problem
- Slowing of Camera Resolution Increases
- HD Analog Rises 2014, Niche Now
- 2015 - Now Rise Cybersecurity
- 2013 - 2017 Rise of PRC Manufacturers
- 2015 - 2017 Race To The Bottom
- 2018 - Now The West vs PRC
- 2019 - Rise AI and Cloud Startups
- 2019 - Now Hostage-as-a-Service Systems
- 2020 - Facial Recognition Negativity
- 2020 - 2022 Coronavirus Impact
- 2021 - Current Now AI and Cloud Mainstream
- 2023 - Rise Direct To Cloud Cameras
2000 - 2005, DVR Era
The first part of the 2000s witnessed the rise of DVRs, replacing VCRs, bringing two important advances - (1) replacing costly and cumbersome VHS tapes with digital recording and (2) enabling monitoring of video surveillance over IP networks.
Recorders were quite expensive in that era, with $5,000 to $10,000 for a 16-channel appliance common, even with limited storage and low resolution (CIF, a fraction of even SD, was widespread). However, it was less expensive than the operational costs of maintaining VHS tapes plus had the benefit that video could be viewed throughout an organization's facilities. Remote viewing over the Internet was possible but given limited bandwidth (max WAN bandwidth of 1Mb/s to 3Mb/s was frequent) and limited CODECs (this was before the rise of H.264) meant that the quality and speed of Internet-based video surveillance watching was poor.
2001, 9/11 Impact
Along with technology rapidly improving in the late 1990s to early 2000s, terrorism in the early 2000s, most notably 9/11, drove increasing demand for video surveillance.
Some of this was good but some bad. On the positive side, it increased awareness and interest in considering emerging technology. And for sellers, it was clearly a boon as the fear of being the next target of terrorism made it easy to justify spending on these systems. On the negative side, many purchases were rashly made on technology that was not mature enough, which resulted in, at best, security theater, and, at worst, a waste of money. This has parallels with the 2020 coronavirus fever camera response.
2006, Infancy IP and VMS
By 2006, the industry was dominated by DVRs and SD analog cameras.
VMS software and IP cameras were still niches. Some megapixel cameras were offered but they were far more expensive than analog ones and only supported MJPEG encoding, making the storage and transmission of these cameras even more expensive.
Analytics was fairly 'hot' in 2006, driven by its potential and VC funding, though with very limited deployments.
The major players were generally Western and Japanese large manufacturers, with PRC China branded sales nearly non-existent in the West (Dahua and Hikvision were mostly unknown) and notable companies today like Axis, Milestone, and Genetec still relative 'startups' (indeed, Avigilon only started selling commercially in 2007).
2008 - 2012, MP Cameras Go H.264
The single biggest driver for IP was the adoption of H.264 for MP cameras. This drove mainstream IP camera deployment and, by extension, VMS software. With MP H.264, IP was able to deliver clear benefits in resolution with reasonable increases in total costs (compared to the earlier MJPEG only MP era).
2009 - 2013, Cloud Hype / Bursts
Along with the rise of MP / IP cameras came a significant interest in connecting those cameras to the cloud. The hope was that it would eliminate on-site recording, on-site maintenance, etc. Bandwidth limitations and poor cloud VMS capabilities doomed this. It never really gained much market share and with EMC dumping Axis, it marked the end of that era / error.
It would take many years for the cloud to re-emerge as a significant player within video surveillance.
2010 - 2018, Struggles For Video Analytics
Video analytics never went mainstream, marred by performance problems and unhappy customers. 2011, with ObjectVideo (OV) suing Bosch, Samsung and Sony confirmed and deepened the problems of video analytics, with OV, one of the most well-funded analytics companies effectively ending commercial sales and suing the industry. OV essentially won, with Avigilon paying nearly $80 million for ObjectVideo patents in 2014. The industry lost, though, as analytics remained a niche offering with minimal industry investment.
2012 - 2014, Rise and Fall of Edge Storage
For a few years, many saw edge storage as being a potential next big thing but it remained a niche. The promise of edge storage was to eliminate NVRs / recorder appliances as the storage and software could be deployed inside the IP camera. Reliability problems hurt early adopters. And the rise of low-cost PRC China NVRs (a few hundred dollars is now commonplace), as well as HD analog for even less (see below), pushed edge storage as more of a niche providing redundancy for higher-end applications.
WDR and Low Light Improvements
Cameras have become much better at handling challenging imaging conditions, especially harsh lighting and darkness. A decade ago, WDR cameras were fairly limited and expensive (this was when Pixim was considered leading edge). Low light performance was generally poor. And these problems were even worse for MP cameras where real WDR was essentially non-existent and low light performance was often terrible. The state of the art in our WDR Shootout 2011 is nothing compared to even 'average' true WDR cameras today.
2015, Smart CODECs Rise
One of the biggest changes in the last 6 years has been the rise of 'Smart CODECs' that regularly delivers 50% bandwidth reductions vs 'un-smart' codecs by dynamically adjusting the compression and I frame intervals based on analyzing the scene. Smart CODECs are independent of H.264 or H.265 and can be used with either. For the first few years of their introduction, they were primarily used with H.264 but are now generally used with H.265 as well.
2018, H.265 Mainstream
While smart codecs reduced the benefit of adding H.265 (by delivering bandwidth savings with 'old' H.264), by 2018, almost all manufacturers were releasing new cameras supporting H.265 + smart codecs.
Storage No Longer Major Problem
While there are certainly cases where storage is still a major challenge, the combination of smart codecs, H.265, and increased hard drive storage / price ratio has made video surveillance storage much less of an issue than ever before. In the 2000s and even the first half of 2010s, storage was a challenge, as much smaller hard drives, less efficient compression, and burgeoning HD resolution demand made the cost and complexity of storage a major factor. Now, storage is generally a simpler matter.
Slowing of Camera Resolution Increases
Contributing to that is camera resolution increases are slowing. In the 2008 - 2013 era, resolution (specifically pixel count) was roughly quadrupling from SD (~0.3MP) to 1.3MP. Now, resolution is continuing to increase, but on average, more in the 3 to 6MP range, which is a much slower rate of increase than in the 2005 - 2015 era. From our resolution usage statistics report, this chart shows the trend over the past 8 years:
While 3MP to 5MP cameras have become the majority, usage of 8MP and higher cameras are still a distinct minority. Moreover, there are very few cameras over 12MP even being sold today and while some talk about 8K (i.e. 33MP), such video surveillance cameras are still mostly conceptual.
2014 HD Analog Rises, Niche Now
While SD analog took a long time to 'die', it was finally killed off by HD analog in ~2015. For more than a decade, IP was the only practical way to deliver MP / HD. But early in the 2010s, HD analog, which transmits over coaxial cable just like NTSC / PAL, emerged. It has wiped out SD analog, becoming a player in home / SMB kit sales and in the low to mid market.
In the past few years, HD analog adoption relative to IP has slowed. While HD analog has increased its maximum resolution to 8MP and has added (limited) Power over Coax, even HD analog manufacturers have favored marketing their IP offerings, mainly relegating HD analog to the most cost-sensitive applications and geographies. See our HD Analog vs IP Guide
2015 - 2020, Rise Cybersecurity
While cybersecurity remains a significant consideration, reduced usage of PRC-manufactured products in light of heightened US and European scrutiny and restrictions has begun to decrease cybersecurity prominence. Previously, notable cybersecurity issues came from easily exploitable backdoors of the industry's largest manufacturers - Dahua backdoor and Hikvision backdoor - with the Dahua backdoor resulting in mass hacks in 2017, as well as critical vulnerabilities in 2021 for both Dahua and Hikvision. IPVM maintains a directory of cybersecurity vulnerabilities for video surveillance products.
The other major element of cybersecurity is the risk of state-sponsored or state-controlled companies. This first emerged as a practical issue in 2016 when Genetec expelled Hikvision and Huawei saying they were security risks. This has certainly increased as the US government has banned the use of those products as well as Dahua.
2013 - 2017, Rise of PRC Manufacturers
Even in 2012, PRC manufacturers had a negligible market share in branded Western sales. For example, see our 2010 Hikvision IP camera test to see how bad they were back then. Indeed, Hikvision saw Western direct branded sales as a 'dream' in 2009.
While PRC manufacturers had, for many years, been OEM suppliers to Western brands, it has only been since 2013 when PRC-branded sales exploded in the West.
Before PRC manufacturers expanded in the Western market, $300 was considered low cost for IP cameras, now $100 (or less) MP cameras are commonplace. In particular, Hikvision has also been very aggressive about offering across the board price cuts monthly, something previously very rare in the industry. These moves combined have resulted in a significant ongoing shift to PRC brands.
2015 - 2017, Race To The Bottom
This led to the 'race to the bottom', as manufacturers kept cutting prices, some to gain share (e.g., Hikvision) and others simply to stay alive.
The race to the bottom has now ended, due to a combination of less effective price cuts with prices having gotten so low, rising local costs as PRC entrants expanded, cybersecurity issues (such as the backdoors), criticism / backlash, and now the US government ban.
2018 To Current, The West vs PRC
In 2018, the US government passed a law to ban US government use and funding for Dahua, Hikvision, and Huawei products. While the primary effect was obviously within the US, this move has increased scrutiny of these PRC manufacturers elsewhere in the West, including the EU Parliament removing Hikvision, citing human rights abuses. This has led to falling Hikvision sales in Europe and North America. See our directory of Hikvision global news reports for more examples.
Moreover, the US government's move to sanction Huawei and repeated reports that Dahua and Hikvision are being considered for sanctions for their billion dollars of contracts in Xinjiang, where a million people are held in concentration camps, could bring further changes to the video surveillance market.
In 2022, further action occurred, including the UK government banning PRC surveillance equipment in sensitive government facilities and the US FCC enacting a prohibition on new product authorizations for Dahua and Hikvision.
The current trend is for increased conflict between the West and the PRC, though shifts are possible in the future.
2019, Rise AI and Cloud Startups
While politics has become a major factor in video surveillance, recently, there has been an emergence of a record number of video surveillance startups.
AI and the cloud drove this. While the previous eras were driven by increases in resolution and price decreases, this era is being far more shaped by software to analyze video using 'deep learning' and to manage video in the cloud. Of course, video analytics and cloud have been around for more than a decade. The difference in 2019 is that a host of companies capitalized on improvements in analytics and the maturity of supporting technologies around the cloud (bandwidth availability, cloud infrastructure, etc.).
For the last 2 decades, open systems have been predominant in video surveillance, whether via NTSC/PAL in analog or ONVIF in IP.
Now, a new wave of entrants, typified by Verkada, Meraki, and Rhombus, have gained traction with closed, locked-in systems, that require buying products and then paying an ongoing subscription for the product to work at all. The largest company in this segment, Verkada, was valued at $3.2 billion after a 2022 venture-backed round.
These companies have seen success so far, adding a fundamental change to how video surveillance is sold and managed. While this fuels the financial results of these providers, we believe this is a net negative for both users and the industry as holding users hostage restricts choice and reduces options in connecting other cameras or systems.
2020 To Current, Fight Facial Recognition
Criticism of facial recognition has increased significantly in many parts of the world, including the US, adding to concerns that many Europeans have had. This criticism over the bias of facial recognition and its misuse has led to calls for its regulation and even for it to be outlawed in an increasing number of areas. Related, The US Fight Over Facial Recognition Explained.
This has caused many facial recognition companies to face challenges, including layoffs at Anyvision (now Oosto) and at FaceFirst. Indeed, Anyvision shifted focus away from video surveillance towards access control to navigate these challenges.
A secondary, but still important issue, is how the wearing of masks has undermined the performance of facial recognition, especially in video surveillance. On the plus side, algorithms are improving and, at some point, mask-wearing will decline, but challenges for facial recognition are significant, for the time being.
2020 To 2022, Coronavirus Impact
As with the entire world, coronavirus significantly impacted video surveillance, including:
- Net revenue declined for the year, as lockdowns hampered buying and installing products. For example, integrators were hit fairly hard but began to recover at yearend.
- A surge in fever camera purchases in mid-2020. The hottest-selling video surveillance item of 2020 was fever cameras. The problem was that many of these devices are rigged and health authorities have increasingly warned about their inefficiency. Despite this, companies who made or sold them did much better financially than those who did not. This trend declined sharply beginning in 2021.
- Cloud systems benefited as the increase in work from home and remote monitoring increased the importance of being able to manage and access systems from anywhere.
As global restrictions continue to ease, pandemic-era fever camera sales have plummeted and supply chain problems have continued to improve. Cloud offerings, however, have continued to gain ground.
2021 to Current, AI and Cloud Mainstream
AI and Cloud have moved beyond niches within video surveillance to become widely adopted by even the largest and most incumbent of competitors. While a few years ago it was commonplace for incumbents to not provide AI or cloud, the opposite is now the case.
The use of AI and the cloud are now fundamentally reshaping how video surveillance is sold, deployed, and used.
2023 Decline of PRC Manufacturers In The West, Challenges Inside of the PRC
While PRC manufacturers surged in the 2010s, the 2020s so far (with the exception of the brief hyper-profitable fever camera boom) have been difficult. FCC restrictions added to the previous NDAA ban on Dahua and Hikvision have put downward pressure on their sales. That plus general tensions with the PRC have made buyers more reticent about buying those products. Moreover, inside the PRC, lockdowns and general economic strain have made it challenging for the once non-stop growth manufacturers to grow (and some have declined, e.g. Dahua Revenue And Profits Fall (Q3 2022), Hikvision Revenue Falls, Profits Plunge Q4 2022 and Q1 2023 Analyzed, RaySharp Sales, Profits Plunge (Q1 2023) Analyzed).
2023 Direct To Cloud Cameras
While cameras that include onboard storage and connect directly to the cloud started in the residential market in the 2010s, this trend is clearly accelerating in the 2020s with various manufacturers now offering such models. The advantage of this approach is simplified installation and support, especially appealing to those who want to limit the cost and complexity of on-site labor. IPVM covers this in-depth in The Strategic Importance of Direct-to-Cloud Cameras.
[Note: this report is periodically reupdated to reflect new history.]
Not to mention true plug 'n play!
Well done JH...simple and concise history lesson.
I still have one of these! Where are those precious tapes?
Nice article! Congrats!
I remember the good 'ol days of using switchers, quad splitters and multiplexers when used with VCRs. This was 1995 to 2002 when PC based DVRs came onto the scene being led by the Koreans - IDIS, Kodicom etc.
Who remembers 'Hyperscan' by Robot (then became sensormatic)? Video transmission over POTS lines.... 1 frame per 10 seconds. Then 8x8 came around and offered a better POTS based video xmission system.
Sanyo by far offered the best VCRs (they actually built VCRs for Sony). I recall selling them by the skid for casino projects along with T160 VHS tapes.
Remember when BNCs came in three pieces, and had to be soldered on?
BNC's ha ha. anybody remember PL-259's and triax cable. Motorola Mocam, Silicon Intensified Target, RCA before Burle and Vicon before Pelco buried them in the dust... adjusting target and beam, manually back focusing lenses. Oops, showing my age.
My dad used to tell me scary stories of burnt-in video tubes when I was a little kid.
And there used to be TV repairmen that made house calls.
They had one or two large boxes containing various replacement tubes. The troubleshooting consisted of swapping-in new tubes until the problem went away, and then swapping back out everything but the last tube replaced.
My next-door neighbor was kind enough to let this pestering 10 year old tag along occasionally, and I was convinced I would soon follow in his trade.
Til we got a Trinitron.
That's when the first rule of electronic repair was "give it a good whack!" which would fix an intermittent connection. Today, it's reboot, then default, then firmware upgrade, then replace, in that order!
Those were the days.
"Percussive maintenance. That'll be $45, please."
Hey, that's how I learned my letters and numbers as a very little kid, running in and out to my dad's van to pull tubes out of the caddy when he was making house calls.Then he had me pull the flat twin lead antenna cables in tight spaced attics.
One of my Chinese partners calls our Industry the "Setting Sun" industry, because it's certainly not a rising sun industry...
Anyone remember plugging a quad or switcher into the 16th port on the DVR because the customer couldn't afford a second DVR but they needed 18, 19 or 20 cams?
Get off my lawn you damn kids!
Note: this report has been updated to cover the past 3 years, most notably the US-China conflict and the rise of AI and cloud startups.
Even for me, it is interesting to look back and see how much has changed in just 3 years.
Fun article! As someone mentioned, need to go back to the 90s when a 16 channel simplex, black and white multiplexor was $5K and a duplex, color mux was $12K! The government bought them by the skid loads and commission rates for independent manufacturer reps was still 9% or 10% and we golfed a lot more! Those were the days! Where is Dedicated Micros today?
Oldest piece of CCTV equipment I know of still in continuous operation is a powered Balun running video down the M6. Built by BT for Pye/Philips in 1969/70.
Some pictures of cameras back then from that guide:
Would love to thank the "UndisclosedDistributor" that attached the cover of the guide that I was part of writing something like 50 years ago.
If anyone would like more history of the very early stages of CCTV ( 1960's +) let me know and I can supply a lot.
Some History -Start of the Hidden Camera Category. Images of the first Covert Smoke Detectors ( SD-100) and the second generation with 4 cameras ( patented).
I kind of hoped you would go all the way back to Germans and V2 :)
Lol, Boris, trying to keep to the time frame we have direct experience :)
What about the rise of the fisheye 360? Ipix ? They bled out making that immersive technology.
I do remember them from the early 2000s. There was certainly a lot of noise and interest back then but they never really became anything.
Today, fisheye is a niche and we cover it in our testing but did not cover it in the history since it's not a particularly big part of the industry.
This article about the history starts in 2000, just a couple years before that the the New York Times published a short op-ed about police cameras in public parks, it has some fascinating quotes when looking back on them from where we are today.
"Even though there is generally no expectation of privacy in a public space, most people expect freedom from government monitoring when they eat lunch on a park bench or stroll down a street. The growing use of police video monitors in New York City may threaten the free and anonymous nature of public space."
"The use of surveillance cameras at [ATM's] may...provide users a sense of safety, but...the installation of cameras at Washington Square Park is a good example of the misapplication of this strategy."
"...there needs to be significant public debate about the wisdom of 24-hour videotaping of lawful movement."
"The right to be let alone includes the right not to be caught on videotape"
Imagine if the ideas espoused by the author of this piece had won out back in 1998...this history would look very different.
We did not include that time frame since those early days did not have a direct impact on the industry at that time, though, Axis had a major impact during the 2006 - 2011 time frame when IP / MP exploded throughout the market.
The report doesn't cover the history of multi-sensor surveillance cameras starting in 2006, and today most manufacturers offer such products. Are you planning to expand the report?
Thanks for the question / comment about multi-sensors. I don't see multi-sensors being so significant that it warrants treatment in a concise history of video surveillance.
When I look at multi-imagers, there certainly a factor in the industry (e.g., we spent months doing a multi-imager shootout) but I don't see them as having fundamentally changed the course of the industry. Maybe it could be addressed as a factor in the decline of PTZs which is another area we did not cover. We'll consider it for the future and feel free to share more thoughts. Thanks!
I'd make the case that if you change the word 'multi-sensor' to 'panoramic', it is fundamental as a transformative impact on the industry. The ability to see your entire security scene differently. A new way to achieve the forever sought 'situational awareness' of your facility. Whether multi-sensor panoramic or de-warped fisheye, from my vantage the Arecont Visions and Sentry360's of 2008-2012 (and the VMS's that integrated their panoramic views as singular cameras) had massive impact on the industry of today.
massive impact on the industry of today
Massive is inherently subjective so I am not going to say if you or I are right.
One massive 'change' that I think we can agree on is the decline in PTZs. 15 years ok, PTZs were very widely used. I don't have stats from that era but it felt like it was 25% or higher (anyone who was tracking such info during that era, please comment). Now, it's clearly just a few percentage of units, at best.
And fisheyes and multi-imagers have definitely been a key factor in that decline though it is not clear how much has simply been the increase in resolution, i.e., NTSC / CIF / 4CIF fixed cameras could barely 'see' any more than right in front of them but megapixel cameras significantly changed that.
Very interesting to see the history, changes, and trends since 2000. AI and cloud processing definitely seem to be the major trends over the next 2-3+ years. What a long way we have come from the VHS and CCTV era!
Excellent job putting this together.
I've been waiting for a good time to bring this up and this discussion seems appropriate.
There is a debate that comes up occasionally around the office regarding the use of "CCTV" versus terms like "I.P. Video", "network video", etc.
I feel there is an sharp distinction between what we knew as analog "CCTV" versus the prominent technology use today. The use of "CCTV", even for modern I.P. systems, is still common outside of the U.S.
I hate to correct colleagues on this but I feel distinction is important. Especially when used in written plans, designs, policies, and standards.
What are your thoughts?
I feel there is an sharp distinction between what we knew as analog "CCTV" versus the prominent technology use today. The use of "CCTV", even for modern I.P. systems, is still common outside of the U.S.
Agreed on both. CCTV literally stands for closed circuit television, which is clearly no longer how this technology operates (i.e., closed-circuit has been replaced by IP video transmission and increasingly to the cloud).
As for the social ramifications of correcting senior people (rarely do people new to the industry call it CCTV) I defer to individual tact and preference.
I just call it "surveillance" nowadays, but then I take my queue from whomever I'm speaking to. I don't care what terminology I have to use, as long as I can make myself understood.
Totally understand your position but as we all know,
When your company monitors your computer remotely for productivity measurement
When you are part of contact tracing for public health
When a nation state implements cell tracking, web intel and dark web monitoring for counter terror or trafficking (people/drugs/weapons etc)
When an "org" monitors your social media posts for security
And so forth, this is all surveillance, so I wonder if we could settle on something specific to video?
i thought we chose a long time ago?
As one of those "Senior People" John referred to.. :-)
As you know, CCTV denotes "Closed Circuit". Historically Triax or Coax based cabling, 1v p to p analog systems. Someone could make an argument that an IP based system, wholly contained within itself, no outside connectivity might also somehow fit the definition but to your point, the question is, should it?
The name of this site, IPVM gives us direction "IPV" but opinions abound. CCTV goes back decades, well beyond the year 2000 back to the 50's 60's when TV station cameras were first used to display medical operations to students so many could see and learn.
Video specifications are still rife with ancient verbiage. So the long and short is, I agree, we need consistency and a standard.
The name of this site, IPVM gives us direction "IPV"
Fun fact, for a few weeks, I called IPVM 'cctvcompany.info' but quickly changed that. In retrospect, good move ;)
I've referred to it as CCTV a few times recently, mainly because the term "CCTV" gets people's brains moving in the right direction quickly. If I say "IP video" or "IP cameras", it's not as abundantly clear. And "video surveillance" just takes so long to say... 😃.
Actually, it's more of a judgement call. If I'm talking to somebody well-versed in IT, then I will be as specific and technically accurate as I can. If I'm talking to somebody who is not tech-savvy, that's when I will call it "CCTV", for simplicity.
For the non-tech-savvy people, they wouldn't know the difference between analog and digital media anyway, so why introduce a new term and potentially cause confusion?
I remember going to the state fair when I was a kid and seeing a live demonstration of a cctv camera. You could look at the camera and see your face on a tv. I went to school the next day to tell all my friends about it and they said I was crazy because that wasn't possible. How far we've come.
Nice review .. and trip down memory lane.
Nice job capturing the last 20 years John!
Update: The report has been updated for the end of 2020 including new sections on facial recognition problems and lock-in systems as well as a revised section on coronavirus.
Very good good information, brings back memories.The first cameras I got involved with was in banks back in the 70;s.Home made unit consisting of a Yaschika half frame film camera driven by a brush less fan motor that had a off center round disk on the shaft that triggered the camera button once every revolution. We got 72 frames and on call out to an activation carry the black bag and development gear. Used the bank vault as a dark room and produced negatives and if required contact print positives on the spot. Next came factory produced cameras that took the whole 330 feet roll of Kodak triax film that we could speed up to 700 asa.Then the Videocon tube came in. Common problems with systems was burn in on monitors, and headlights.Recall one time where the camera failed without and warning and it showed a prison cell empty when the prisoner was in. Just about got the hang of repairing these when the Newvicon tube came in, tube cost the same as a complete Videocon camera.Next the bucket brigade area started and I happened to be the proud owner of a 75Baud modem and free internet thanks the the Uni.Amazing world.
This is great lesson as it followed all the happenings in world of surveillance
This is updated for 2023, with new sections on challenges for PRC manufacturers and rise of direct to cloud offerings.
Great article and a timeline for the newcomers or old-timers who have been away...;)
My only two-cents here is the penny-stock market is full with "startups" that tout AI as their key technology. It is by far the biggest hype, bigger than the cloud ever was.
It has become a meaningless term used for OCR open-source algo as much as for more sophisticated pixel-change algo's. I'd bet that most companies in the industry don't use a real AI backbone.