Are you making it? Video surveillance will be the world's biggest data generator

I remember watching a movie a few years ago, Smith starring in the "Enemy of the State", the film in every corner of the United States is full of cameras, the CIA can call these cameras at any time to track, countless cameras formed a huge network ... It was a bit of sci-fi and a little back-cold.

But in a few short years, this has become a reality, millions of cameras have been networked, to our people's financial security, transportation has brought a lot of convenience, but this is only the beginning, in fact, now the technology has been able to easily intelligent recognition, people count and even behavior recognition, This huge video data analysis is closer to the real world than traditional internet data, what exactly does this Pandora's box bring?

Video surveillance will be the world's largest data generator

In the 2012 the world added 2.8ZB of data, equivalent to 3 billion TB. If the video data is stored, the data will double.

Take a domestic province for example, public video surveillance has exceeded the 100多万个 camera, the total volume of nearly 4 million, to estimate that the number of cameras in the country is not less than 40 million, an investment bank reported that the industry is still increasing by 20% per annum, the industry leader of the company's 12 annual report revealed sales volume of 5.7 million sets ( With front and rear end), increased by 37%. The other figure is staggering, with the number of cameras in the UK reaching 1:15 per capita. Roughly the world has at least 100 million cameras in the corner quietly watching.

If this data is stored, it will be super large data. Assuming that all cameras are switched to 1080P, the amount of data per day is 64G, the annual data volume is 23T, the world will produce 2.3 billion t of video data every year.

This is a terrible number. I do not know is fortunate or unfortunate, the vast majority of video surveillance data storage will not be more than a day, because from the current view, compared with the price of storage, these data are worthless.

The emerging super video surveillance cloud

Video surveillance is no longer the way it used to be, with several cameras connected to the building's security room to get a glimpse of the mysterious room. In fact, the surveillance camera has become an intelligent terminal, there is a network, can control even two-way audio, camera interconnection has become a trend.

such as public video, has been from the interconnection of a city, developed to a provincial interconnection, and gradually to a national interconnection, whether the new webcam, or the old analog camera +dvs/dvr, are beginning to accelerate access to a large cloud platform, from simple video streaming media data to the PTZ operation, Each terminal becomes an intelligent terminal that can be controlled remotely. At the other end of the spectrum, the data is packaged into public use and opened to the public to see how many mobile phones are flooded in the media to see the news of urban road traffic videos in real time.

In addition to public video, countless private videos are also forming private clouds, branching out into huge chains of supermarkets, chain hotels, logistics companies, financial institutions and even industrial production lines, and countless endpoints are converging on the cloud. And many countries also require private video data to be directed open, can be invoked and manipulated.

This is a different cloud network from the Internet, from perception to interconnection to application, but this perception is to see the world with countless eyes.

Welcome to the "Truman World".

The value of large video data is still untapped.

In general, there are too many redundant video data, only the unexpected events will be back to look at the record, so most of the video data storage is very short, more than a week of few. On the other hand, because the data read and write requirements high, cloud storage is extremely expensive (thousands of Yuan 1TB), so most of the data are there front-end.

Video Intelligent analysis of the original intention is to solve the problem of redundant data, people from the boring monitoring to liberate, to help people better eye view. The core of the so-called intelligence is target recognition and behavior analysis, according to the requirements of the application set rules, when the information in the video to comply with the rules before the alarm and record, common applications include intrusion detection, perimeter alarm, vehicle identification, traffic violations monitoring.

The current industry mainstream intelligent analysis manufacturers, such as foreign Objectvideo, Ioimage, Emza, the domestic technologies technology, Wenan technology, intellectual state technology, and so on, are basically around the "monitoring" word, belong to the primary processing of video data, is generally just a single camera video information for real-time processing, According to certain rules to produce event data (such as alarm event, vehicle count), across space and time range of two processing analysis is still relatively few, therefore, the video surveillance field does not really form the internet sense of large data.

But video surveillance data must be a treasure trove of the next big data. On the one hand, it has a typical large data 4V characteristics, large amount of data, diversification, seemingly disorderly, but implicit in countless people and things behavior. On the other hand, it is the portrayal of the real world, which is very different from the large data obtained by the Internet, the real world contains countless information that is difficult to express in the form of words, such as people through vision can quickly form a judgment, a place is prosperity or recession, the atmosphere is tense or cheerful.

Of course, the premise is that storage costs can be reduced, data processing capacity can rise.

Great potential applications

For example, for a shopping mall, in addition to the need for security, the video data mining two times, you can collect the customer's sex, age, wearing information, you can count the customer's shopping path, stay mode, gather hot spots, or even two or more times, to the store cycle. This is similar to web site visit analysis, can provide basic data for the optimization of the mall. If it is a chain-type enterprise, but also can combine the data of many stores, access to regional and even national data.

Also magnified to a city, the pedestrian data of every street, flow data are in which, the whole collection of these data, is a city of human car distribution map, static, such as the characteristics of different areas of people, car characteristics, dynamic, such as the path of human vehicles, stay mode, which is very valuable for urban planning and management.

If you combine it with the timeline, the information gathered from countless cameras, can also see a country, a region, a city changes, as the "big data" said, can even predict the trend, such as whether more shops in the new renovation opened or more closures, is more restaurants or more clothing stores, These changes together we can see the demographic changes, economic trends, trends, changes in the natural environment and even the degree of human happiness and tension.

This is not a myth, now the technology has been completely done. On the one hand, the popularity of high-definition cameras, video information quality is better; On the other hand, the level of intelligent analysis has been quite high, the identification and separation of objects, the recognition of the face, the identification of color words and figures, the analysis of the changes in the object and even the monitoring of violence.

But to realize data mining in this sense, the need for a large number of metadata records, and even unrelated to the monitoring of metadata, to multidimensional analysis, the need for massive data collection storage and large-scale processing, and finally need to be based on the location and time of the analysis of the relevance of integration, This huge resource and cost is an obstacle to opening the door to big data.

Conflict between technology and human nature

A week ago Google announced that it would prohibit glass apps from using face recognition until the effective privacy protection mechanism was perfected. Previously, Google has been questioned by lawmakers for Glass's privacy policy, including questions about whether facial recognition violates user privacy, whether Google Glass collects user information, and how Google protects user data.

Video surveillance also involves privacy issues. The technology of face recognition in video surveillance has been quite mature, if the storage is big enough, the processing is strong enough, in countless cloud data in the face search can also produce very many meaningful applications, such as searching for lost children, in the future only need to input photos, can be a wide range of intelligent search, which is more effective than relying on tactics.

But this invasion of privacy is difficult to bypass the pain, domestic legislation in this area is still blank, such as the previous microblog burst of "touch milk door" is a typical civil privacy violations of the incident. In the supervision of the need to be clear who can build a camera, where can be built, how to save data, how to protect information is not leaking.

But in reality, the rapid growth of the camera has shown that citizen privacy or public management, but with the formation of video surveillance cloud and to the cloud storage, the data is more powerful, through cloud data, can even describe a person's life track, hobbies, ta like to go to which shops, where activities, Dress hobby, often with who together ... Only God can see, the clouds are looking at you.

Technological progress will always bring new challenges, the future wise people must come up with privacy protection methods, only your authorization to see your presence in the video, or, our three-view evolution, adapted to the development of technology.

The world is a stage, God is smiling in the clouds.