Security cameras with Artificial Intelligence are not only used to improve the security of homes or businesses. There are multiple utilities that can be applied to this and they also depend a lot on the creativity of the individual who uses them. For example, they can be used to control the capacity of a nightclub, to detect the use of certain objects, such as weapons or even to identify certain items of clothing. In addition to this, they can also be used for more common things, such as differentiating a person from an animal and thus avoiding a false alarm.
Technologies and algorithms used in video surveillance.
Before talking about the algorithms used, it should be remembered that this branch is Computer Vision, so when looking for algorithms and techniques, it is advisable to focus on libraries and technologies that are specifically made for Computer Vision.
- OpenCV: is one of the most famous libraries for image classification. With this library, it is possible to identify types of objects, types of living beings or counting gauges. This library usually uses simpler Machine Learning algorithms than other technologies, so for example, for facial recognition it is more advisable to use more complex libraries and architectures that are not included in the OpenCV package.
- Tensorflow: this software package already uses advanced neural network algorithms, so it increases the accuracy of facial recognition and identifies situations with greater precision. It should also be noted that it requires more resources and better hardware.
- DenseNet121: is a convolutional neural network architecture that works very well for image classification. That is, real-time image classification algorithms are used to identify and work with pattern identification. This neural network architecture was invented in 2017, so it is not so old, compared to other algorithms and mathematical calculations.
- PyTorch: this artificial intelligence library is employed in multiple areas of Artificial Intelligence and one of them is Computer Vision. It is possible to use it to detect anomalies, to track objects in real time or to improve the use of other libraries such as OpenCV, combining them in the same software.
As can be seen, most of the popular AI libraries are usually the same as those used in other branches, except OpenCV and some others. This shows that there is still much to be developed in this branch of AI, so that in the future there will be much more specific libraries.
How to put Artificial Intelligence to an old installation?
It is not necessary to change the entire installation to be able to use AI in a video surveillance system, however, it is possible that some low resolution cameras may not be able to apply it as accurately or that a computer with a high graphics power and RAM memory is needed to process the video and sound coming from those cameras.
Below is a list of AI programs that can be used in video surveillance systems.
AI software for video surveillance installations.
Veesion.
This software is capable of detecting thefts in supermarkets, thanks to Artificial Intelligence. Every month they detect more than 100,000 theft attempts and work with companies such as Carrefour or Suma. Besides being able to be used in common stores, it is also useful for pharmacies, where there are high-cost products. They have conducted studies and have concluded that it can prevent up to 60% of thefts.
AcuSense.
This Hikvision software is specially made to avoid false alarms, however it also has other functions such as setting perimeters or sending images to mobile devices, at the time of the detected anomaly. It uses Deep Learning for its operation.
Scati.
It is a software that is able to manage capacity, queues, prevent theft and more use cases, thanks to the use of the implementation of AI in video surveillance systems. For example, they offer their services to hotels, arcades and banks. This is just one example, as they have solutions for many types of businesses.
The best security camera companies.
If we base on market share, the best security camera brands are the following. Since they have a higher number of sales and a greater diversity to choose from. Especially if you are looking for special features, such as the incorporation of object detection or artificial intelligence techniques.
Axis Communications.
This company is based in Sweden and is very popular, offering applications for its security cameras with functions such as facial recognition, moving object detection, audio analysis or behavioral analysis. As far as audio analysis is concerned, many people may not have realized that audio is another option to work with, for example when hearing gunshots or screams.
Regarding interesting information about this company, we can highlight that they have more than 4,000 employees, that it is the favorite brand of many governments and security forces, and that it belongs to Canon Group, so it now belongs to a Japanese company, which by the way, is listed on the stock exchange.
Hikvision.
Hikvision uses Deep Learning algorithms to apply AI to video surveillance. For example, it is able to detect cats and not set off alarms. It also differentiates the different areas in which a human is located and can trigger warnings when they approach a certain monitored area, for example, if a child approaches a dangerous place. A real-world example that Hikvision has employed, is using Deep Learning in a housing estate in South Africa. Due to increased violence and the fact that it is a luxury housing estate, it monitors in real time, even during the night, if any intruders are approaching the premises. So they have achieved greater efficiency and savings than with a conventional system.
Hikvision is the company with the world’s largest market share in the security camera sector and is headquartered in China, as well as being listed on the stock market. However, being a Chinese company, special care must be taken with currency exchange and its variation.
Bosch Security Systems.
Bosch is a German company that also has a fairly sizable market share and is also applying Artificial Intelligence in its security cameras. Some of the real use cases they are applying are: safer railway crossings, fire detection by camera, detection of emergency exit blockages, detection of badly parked cars, detection of abandoned objects, people monitoring, traffic monitoring and many more uses, which you can see in their official website, see real examples of AI with cameras. They call this technology Video Analytics and it is already integrated in many of their systems.
This company has been working since 1886, being one of the most relevant companies in the European Union, offering its development in branches such as the energy, railway, industrial, security and mobility industries. They are even involved in blockchain projects as well.
Dahua Technology.
This Chinese company, is the second largest in the world, when it comes to video surveillance. Regarding Artificial Intelligence, they are using it in their “WizSense” technology service which is able to do facial recognition, monitor perimeters, such as that children do not pass into a dangerous area like a swimming pool or participants of a party, into a restricted area. It also ensures that humans are not mistaken for objects or animals. In addition to improving the quality of the image, to provide security during the night.
Something very interesting about this company is that they sell a product to open doors with facial recognition or access to certain areas. It is a kind of small screen that is attached to a wall or door, with a small camera and keyboard. Another interesting product is their drones with Artificial Intelligence, which can also be used for surveillance.
Panasonic.
The famous Japanese company, Panasonic, has effectively specialized in offering AI in its security cameras. They offer multiple uses, some of them are: security for museums, for example that attendees do not touch the works of art; occupancy detection in rooms and stores; detection of violations on the road and vehicle detection; detection of door locking, also facial detection accurately. Deep Learning algorithms are used for all these functions. Something to note about Panasonic, is that they offer a very high manufacturing quality, like most Japanese companies and also noteworthy, that high resolution lenses are used, as it is a brand that owns patents and pioneer in the audiovisual sector.
They are also applying Artificial Intelligence in engineering, as with the TOUGHBOOK device, which will be able to use AI software, to streamline and improve the safety of the work of engineers in the civil engineering sector.
Xiaomi.
This company is applying AI in cameras in a more original or different way, as it markets small security cameras, intended for domestic areas, that use AI algorithms for functions such as voice recognition or notifications based on camera records, such as gestures or object detection. They are also marketing devices that can be connected to video surveillance cameras, such as door locks that send signals to the camera and storage of anomalous events in cloud servers, so that those videos cannot be physically erased, thus preventing hard drive theft or provoked disappearances. As you can see, Xiaomi is revolutionizing technology in several industries at once and also in an original and cutting-edge way.
Apply creativity to increase use cases.
As you can see in many of the examples above, creativity is essential to be able to make this technology useful, so if you are a creative person you can also create very profitable programs for future customers or specialize in a particular use case. Some examples of very creative uses:
- Detecting weapons: this could be very effective in combating terrorism by governments or preventing robberies where a high level of violence is used. Thus achieving a safer society thanks Artificial Intelligence.
- Avoid robberies at ATMs: banks can analyze strange behaviors, thanks to the use of this type of programs.
- Detect infidelity: it is not a very ethical way, however for domestic use, it would not be unreasonable for someone to use the people counting function to count more than one adult person in the same room, such as the bedroom.
- Prevent drug trafficking: with cameras that analyze anomalous behavior in certain places.
- Preventing criminals from approaching schools and kindergartens: thanks to facial recognition, this could be possible.
- Preventing accidents: stopping a train when it receives an alarm that someone has jumped onto the tracks.
Do you have any original ideas? Leave a comment.
My personal opinion.
I think it is an excellent branch to specialize in and also, unfortunately, it is possible that some governments use it to restrict the freedoms of citizens. However, with good laws that protect people’s privacy, it is possible that thanks to this technology many crimes, accidents and antisocial behaviors can be avoided.