Why do CCTV developers love retail more than manufacturing?
Did you know that in the range of software vendors for video systems there are many solutions for the applied tasks of trading - modules for counting visitors, determining the length of the queue, monitoring operations at the checkout, etc. And there are practically no proposals for solving production and industrial problems. That's because we, the developers of video surveillance software, for production, unlike retail, rarely develop, and it costs a lot.
Why? Let's get it right.
Initially, CCTV systems were invented for security. They solved traditional tasks: storing the archive, displaying on operator screens, motion detection and searching the archive. Regardless of which industry the object on which the video system is installed belongs to, the approach to solving these problems is the same. And the technologies for their solutions, developed once, are successfully applied at millions of different objects.
Over time, video surveillance systems went further and learned to solve somewhat more complex problems - to analyze video: recognize faces, recognize car numbers, and search by various parameters. And again, no matter what the object is, the process is approximately the same, because, for example, there is no difference for the license plate recognition module, there is a car at the gates of a shopping center or factory.
The next step is to solve narrower specialized tasks that go beyond security. First of all, such systems appeared for retail. They can analyze the movement of customers, determine the length of the queue, count visitors, identify the most active areas of the trading floor. And all this works quite effectively.
Macroscop traffic heatmap in the lobby of the mall
Now it’s logical to take the next step: if we taught systems to analyze video and solve applied problems at the same objects, why not introduce it everywhere? Not only in trade, but, for example, in production, and then systems can replace workers who are engaged in low-skilled labor.
In practice, everything is not so simple. If the basic tasks of ensuring security are the same, the tasks of video analysis for retail are also the same (because it is not so important what to sell, stores as a whole sell the same, regardless of what is on the shelves), then the production has a wide variety of tasks. All production is different and each has its own production process.
Judge for yourself. At Macroscop, we receive many video analytics requests from various manufacturing companies. For example, with the help of video systems, our customers want to:
• recognize a broken tooth of an excavator;
• determine fractions of crushed stone in the body of dump trucks at a quarry;
• count plastic bottles in a pallet;
• detect gems on a conveyor belt.
Even these tasks are so different and specific, that for each of them, it is necessary to engage in individual development. Standard tools developed once, they can not be solved. And such an individual development is expensive.
It is logical that a widely replicated product is cheaper than a product that is produced in smaller volumes. In order to develop technology and establish production, it is necessary to make certain investments, which should pay off through sales. And the more copies of the developed product are sold, the smaller the amount of production costs is invested in the cost of each unit. When your task is completely individual, the product for its solution is produced in a single copy (and the likelihood that someone else will buy it sometime is small), all investments of developers are also included in its price. The proportion is simple: lower demand - higher price.
Macroscop also has examples of self-solving non-standard production tasks at minimal cost, when customers simply were smart and used standard video analysis tools.
For example, one of the companies involved in the production of roofing materials solves the problem of detecting defects (holes) in the produced material with the help of a motion detector using the video system. Here is what their system is like:
Material moves along a conveyor belt through a dark box. At the bottom of the box, a light source is installed, directed vertically upward (through the tape), the video camera is located on top, opposite the source, and directed vertically downward. If there are no holes in the material segment, the picture on the camera is absolutely black, but if there is a defect in the material section in the dark box, the camera captures the gaps (spots of light). Macroscop is configured to react to the appearance of these light spots (using a motion detector), and upon the event an external application is launched, which informs the operator about the presence of a defect.
The cost of the software component of this solution is 1800 rubles (per 1 camera).
Not everyone shares our point of view. Some of our colleagues in the workshop have the opinion that, in general, all tasks, no matter how different they may be, can be classified and summarized for some universal development.
But it is possible to classify effectively only when really close problems are found in the same classes, for the solution of which the same approaches can be applied. If in one class there were 5 seemingly similar problems, but there is still no universal algorithm for solving them, this classification becomes meaningless. To which class, for example, does the task of detecting a broken tooth of an excavator belong?
On the other hand, it is clear that the future is precisely for this. After all, you can classify tasks, for example, to recognize some objects or some event. And, probably, very soon algorithms will appear in the video systems that can recognize anything. To some extent, such developments are applicable now.
Not so long ago we met blippar, which, as she herself says, made the world's first visual browser. They made an application for mobile devices that recognizes any objects that fell into the lens of the camera of a mobile phone, and displays content about them. Already today this is a very working application, which shows quite good results.
All this suggests that promising video analysis technologies are already being created, which are universal, work in different conditions and do not require refinement for specific tasks.
It is quite obvious that in the future the recognition problem will become a fundamental, basic function that can be applied everywhere, including at production facilities for solving special problems.
But you need to separate the present and the future. And today, while the technologies are not at such a level, customers with a narrow specific production task have three options:
1. Pay quite a lot of money for individual development.
2. Be smart and resourceful, using standard tools in non-standard ways.
3. Or not use video analysis, but connect human resources.
Why? Let's get it right.
Initially, CCTV systems were invented for security. They solved traditional tasks: storing the archive, displaying on operator screens, motion detection and searching the archive. Regardless of which industry the object on which the video system is installed belongs to, the approach to solving these problems is the same. And the technologies for their solutions, developed once, are successfully applied at millions of different objects.
Over time, video surveillance systems went further and learned to solve somewhat more complex problems - to analyze video: recognize faces, recognize car numbers, and search by various parameters. And again, no matter what the object is, the process is approximately the same, because, for example, there is no difference for the license plate recognition module, there is a car at the gates of a shopping center or factory.
The next step is to solve narrower specialized tasks that go beyond security. First of all, such systems appeared for retail. They can analyze the movement of customers, determine the length of the queue, count visitors, identify the most active areas of the trading floor. And all this works quite effectively.
Macroscop traffic heatmap in the lobby of the mall
Now it’s logical to take the next step: if we taught systems to analyze video and solve applied problems at the same objects, why not introduce it everywhere? Not only in trade, but, for example, in production, and then systems can replace workers who are engaged in low-skilled labor.
There it was ...
In practice, everything is not so simple. If the basic tasks of ensuring security are the same, the tasks of video analysis for retail are also the same (because it is not so important what to sell, stores as a whole sell the same, regardless of what is on the shelves), then the production has a wide variety of tasks. All production is different and each has its own production process.
Judge for yourself. At Macroscop, we receive many video analytics requests from various manufacturing companies. For example, with the help of video systems, our customers want to:
• recognize a broken tooth of an excavator;
• determine fractions of crushed stone in the body of dump trucks at a quarry;
• count plastic bottles in a pallet;
• detect gems on a conveyor belt.
Even these tasks are so different and specific, that for each of them, it is necessary to engage in individual development. Standard tools developed once, they can not be solved. And such an individual development is expensive.
It is logical that a widely replicated product is cheaper than a product that is produced in smaller volumes. In order to develop technology and establish production, it is necessary to make certain investments, which should pay off through sales. And the more copies of the developed product are sold, the smaller the amount of production costs is invested in the cost of each unit. When your task is completely individual, the product for its solution is produced in a single copy (and the likelihood that someone else will buy it sometime is small), all investments of developers are also included in its price. The proportion is simple: lower demand - higher price.
Can save
Macroscop also has examples of self-solving non-standard production tasks at minimal cost, when customers simply were smart and used standard video analysis tools.
Life hack
For example, one of the companies involved in the production of roofing materials solves the problem of detecting defects (holes) in the produced material with the help of a motion detector using the video system. Here is what their system is like:
Material moves along a conveyor belt through a dark box. At the bottom of the box, a light source is installed, directed vertically upward (through the tape), the video camera is located on top, opposite the source, and directed vertically downward. If there are no holes in the material segment, the picture on the camera is absolutely black, but if there is a defect in the material section in the dark box, the camera captures the gaps (spots of light). Macroscop is configured to react to the appearance of these light spots (using a motion detector), and upon the event an external application is launched, which informs the operator about the presence of a defect.
The cost of the software component of this solution is 1800 rubles (per 1 camera).
Alternative point of view
Not everyone shares our point of view. Some of our colleagues in the workshop have the opinion that, in general, all tasks, no matter how different they may be, can be classified and summarized for some universal development.
But it is possible to classify effectively only when really close problems are found in the same classes, for the solution of which the same approaches can be applied. If in one class there were 5 seemingly similar problems, but there is still no universal algorithm for solving them, this classification becomes meaningless. To which class, for example, does the task of detecting a broken tooth of an excavator belong?
On the other hand, it is clear that the future is precisely for this. After all, you can classify tasks, for example, to recognize some objects or some event. And, probably, very soon algorithms will appear in the video systems that can recognize anything. To some extent, such developments are applicable now.
Not so long ago we met blippar, which, as she herself says, made the world's first visual browser. They made an application for mobile devices that recognizes any objects that fell into the lens of the camera of a mobile phone, and displays content about them. Already today this is a very working application, which shows quite good results.
Eventually
All this suggests that promising video analysis technologies are already being created, which are universal, work in different conditions and do not require refinement for specific tasks.
It is quite obvious that in the future the recognition problem will become a fundamental, basic function that can be applied everywhere, including at production facilities for solving special problems.
But you need to separate the present and the future. And today, while the technologies are not at such a level, customers with a narrow specific production task have three options:
1. Pay quite a lot of money for individual development.
2. Be smart and resourceful, using standard tools in non-standard ways.
3. Or not use video analysis, but connect human resources.