How to use video analytics and metadata to prevent terrorist attacks

https://www.sourcesecurity.com/news/articles/video-analytics-metadata-prevent-terrorist-attacks-co-6163-ga.23645.html

Today ‘terrorism’ has become a word we use and hear every day. The goal of terrorism is a media product - information delivered to nearly every house in the world. So, the weapon of terrorism is information.
Therefore, the way we defend and prevent terrorism must also be based on intelligent processing of information - and an early awareness of potential threats and effective preventive action may eliminate most attacks. Video analytics, automated surveillance and AI decision-making are going to change the rules of the struggle between civilians and terrorists by making attempted attacks predictable, senseless and silent. In this article, we will evaluate to what extent technology can investigate and prevent terror crimes considering the latest technology innovations.
Civilian feedback helps terrorists to accomplish missionIn order to achieve their main goal - loud media response - terrorists and those who order the attacks use unpredictable tactics and the element of surprise; so that after every attack, the media discusses for months the circumstances and their insanity.
Unfortunately, each time it happens our society seems to be unprepared. As the media environment grows, terror attacks attract more attention, and the feedback of civilians actually helps the terrorists to accomplish their mission.
Features of terrorist crimesCounter-terrorist specialists highlight, among the others, the following inherent symptoms of terror crimes:

• Unpredictability
• Public visibility
• Enormous social resonance

The question is: Are there technological solutions that could treat these symptoms at a low level? Crime investigations are based on objective indisputable facts that can be used against suspects in a court.
The facts are:

• Video surveillance materials
• Facial recognition and ANPR metadata
• Audio data (e.g. phone calls)
• Internet communication logs
• Other registered human actions
  
  

Metadata sources and analytical systemsTo be able to collect and analyse that data, it needs to be in a data format that an analytical system will be able to process. Metadata can be generated by processing data of the above sources. Metadata can be stored in relational databases or in blockchain, so it can be a reference for an analytical system or law enforcement structures.


​

Aggregation of metadata sources could be constructive because it would significantly increase metadata availability for analytical systems and will improve metadata quality. This would surely require replacement of most of existing security systems and standardisation of new systems so to ensure maximal compatibility of metadata sources and analytical systems.
Offline video analyticsAs these improvements are difficult to develop and implement globally, replacement solutions are being offered currently in the security market. One of them is the concept of offline video analytics, which generates and analyses metadata from any video source. Video sources may vary from ‘old school’ analogue cameras to high-resolution IP cameras recorded in any digital format. Quality of the metadata generated from offline analytical systems is almost unaffected.
High quality metadata can be analysed and investigated automatically or semi-automatically for violations, crimes and terror activity. Automatic or semi-automatic investigation can be based on crime scenarios, behaviour patterns, forensic search, face and vehicle recognition and synoptical search. Fast and effective investigation of terror activities may prevent attacks and also can reduce the number of active terrorists.


​Deep learning and neural network technologiesHowever, realtime crime and terror prevention requires instant metadata generation and analysis. The investigation instruments mentioned above would not be of the same efficiency. Firstly, processing capabilities of analytical system must be very high because the system should be able to record data, generate metadata and analyse it at the same time in realtime conditions. Currently the most powerful server processors can run only tens of detectors so it becomes very costly. That is why these kinds of solutions are only used in critical infrastructure. However, if they were used widely they would dramatically reduce the number of criminal and terror activities.
Deep learning and neural network technologies (so-called artificial intelligence - AI) are coming to the security market to replace classic video analytics. These systems are not yet much more efficient hardware-wise; however, they have greater potential and they are cheaper. Behaviour patterns, actions, sounds, speech, faces, car number plates and many other metadata types can be identified and collected and analysed by AI in realtime.


​Emotion recognition/vibraimage technologyEmotion recognition (or vibraimage) technology measures micromovements (vibration) of a person by processing video from a camera or any video source. Human head microvibration is linked with the vestibular-emotional reflex (VER) and depends on emotional status. Vibraimage systems detect human emotions by the control of 3D head-neck movements accumulated in several frames of video processing. Vibraimage is a system that detects all human emotions.
Blockchain can bring awareness of different views. Imagine if the security surveillance and analytical AI system knew about each person’s life and social background so it could make automatic decisions to give more surveillance priority to those who potentially could take negative action.
Although security equipment is becoming more affordable, the budgeting of security systems at a government and private level is still the biggest problem. As the global population is growing and migration is getting more intense, public and private security is becoming a natural need. Meanwhile, the security market is ready to deliver solutions that can instantly investigate and even prevent terror activities.