Surveillance is basically watching over the people for their own protection. According to Wikipedia, Surveillance is the monitoring of behaviour, many activities, or information for the purpose of information gathering, influencing, managing or directing. This can include observation from a distance by means of electronic equipment, such as closed-circuit Television (CCTV), or interception of electronically transmitted information like Internet traffic. Increasingly, governments may also obtain consumer data through the purchase of online information, effectively expanding surveillance capabilities through commercially available digital records. It can also include simple technical methods, such as human intelligence gathering and postal interception.
In recent years, significant advancements have happened in the surveillance field due to rapid technological developments in Artificial Intelligence and its subfields. One of the major subfields of AI that directly has an impact on surveillance is Computer vision. Computer vision is basically how computers can understand their surroundings with visual inputs and self-decision-making processes. I have written an article on basic introduction to computer vision in here.
using CV technology and Machine learning algorithms, the modern surveillance is changing and becoming more and more advanced technological field.
The previous figure shows how a detection algorithm can detect faces in real-time (Live) in a given surveillance camera. the model can scan the video frame by frame by using image processing and detect faces and track them for given period of time. This model can be used as a primary tool for surveillance using CCTV in specific area/site. Just like this simple example model, the computer vision and other subfields of AI can be used to build advanced tools for modern surveillance.
This figure shows how machine learning algorithms and computer vision can be used to build facial structure analyse visual inputs and build feature-prediction or extraction tools. The following figure shows how a more advanced model can predict 3 important features; age, gender, and facial emotion at the moment more accurately.
These feature prediction or extraction from facial analysing is essential areas in modern surveillance.
Previous figure is a Screenshot from sample surveillance video. When the proposed video go through deep machine learning algorithm, the neural network is working on finding similar features for clustering of faces under specific gender and age range. This is very important for identifying certain individuals/persons of interest, identify registered criminals, terrorists, on any CCTV camera footage in real time and make alerts. this is a prime example for using AI and computer vision for advanced surveillance.
Identifying guns and weapons in real time is another major security tool for advanced surveillance system. This can be used to create more safer environment for public and react in real time to prevent major disasters.
Security cameras and video surveillance systems have become important infrastructures for ensuring the safety and security of the general public. However, the detection of high-risk situations through these systems is still performed manually in many cities. The lack of manpower in the security sector and limited performance of human may result in undetected dangers or delays in detecting threats, posing risks for the public. In response, various parties have developed real-time and automated solutions for identifying risks based on surveillance videos.
In 2019, there was a major terrorist attack in Sri Lanka. Many innocent people dead because of it. This was an eye opening moment for me. Then I saw many other terror attacks and mass murders against innocent civilians around the world. The latest moment was mass knife attack on the Australia where lived at the time. So, I realized protecting innocent civilians from evil is a priority of rapidly Advancing AI and robotics.
There are many places where the crime rate caused by firearms or knives is very high, especially in places where they are allowed. The early detection of potentially violent situations is of paramount importance for citizens’ security. One way to prevent these situations is by detecting the presence of dangerous objects such as handguns and knives in surveillance videos. So, the current surveillance and control systems still require human monitoring and intervention. To prevent this, the early detection of these weapons using deep learning techniques through video security in real time and minimizing the computational burden.
Using AI-powered tools and hardware devices is a great solution for this problem. AI have many advantages compare to other protection method including accuracy, can work without getting tired, faster computations, cheaper cost, etc. modern AI and its subfields can be used to build several advanced surveillance applications including:
- Weapon detection
- anomaly / abnormal activity detection
- advanced Facial recognition
- Security alert management
- AI-based security hardware operation
- real time responding
- event management tools
- parking management
It is clear that using AI and Computer Vision will play a key role in modern and future surveillance systems. So, it is important to do continuous research and develop more and more advanced systems with the sole purpose of protecting innocent civilians from harmful activities.