Transport security can be revolutionized through the power of artificial intelligence (AI) and anomaly detection, writes Ken LaMarca, CEO of Active Intelligence.
Transportation facilities serve as the central nervous system for modern cities, providing indispensable infrastructure for the daily lives of residents, commuters and visitors.
These dynamic hubs, essential in the seamless movement of people and goods, inevitably face evolving security challenges due to their public nature and often limited security screening measures.
While surveillance systems employing traditional video analytics solutions have been deployed to solve these challenges, they come with limitations and complex setup requirements.
Recent advancements in artificial intelligence (AI) and machine learning (ML) technologies have ushered in a new era of security solutions, offering the potential to effectively address and alleviate many of these challenges.
In this article we’ll dive into three of the biggest transportation security challenges and showcase how integrating AI-powered anomaly detection into transportation video security operations can address them while improving safety and security.
Transportation facilities, by design, witness a constant influx of vehicles and pedestrians.
While essential for their functionality, this high traffic volume poses unique challenges for security operations.
Ensuring the safety of passengers within the dynamic environment of transportation facilities, especially those open to the general public 24/7/365, is no small feat.
While video analytic solutions exist to spot specific scenarios within a transit hub, like an abandoned bag or parcel, this environment can breed a literally endless variety of unwanted or unplanned behavior, with threats evolving and changing over time.
To detect these changing threats, adjustments of video analytics can range from a minor tweak to a complete reconfiguration. In many instances, there may not be an analytic solution with the capability to detect this new threat.
What’s more, many analytics are only helpful when conducting a forensic investigation of an incident after it has occurred.
In this case, detecting events in real time requires dedicating a security operator to be singularly focused on just a few cameras – an expensive and unrealistic proposition.
Additionally, the attention span and focus of security operators is limited, even among the best in the industry – it’s just human nature.
It would be unfair to expect an operator to be focused on monitoring cameras every second of their shift.
Add to this the fact that transit hubs often have hundreds of cameras to be monitored, covering the constant activity of the thousands of people passing through and only a few staff members to watch them.
AI-powered anomaly detection is not just another video analytic – in fact, it is not an analytic at all.
It picks up where traditional video analytics fall short, helping to address these challenges and go even further to improve safety and security by detecting anomalous events in real time.
AI allows anomaly detection solutions to constantly improve, allowing them to detect most irregularities in real-time while simultaneously learning and improving detection capability.
Each camera view gets its own uniquely tuned AI model, allowing for more accurate identification in varying environments.
Real-time detection means your team can address unwanted behavior or emergencies immediately, reducing risk to occupants as soon as possible, rather than performing damage control once an incident has occurred.
Certain offerings can be rapidly spun up on site and begin detecting anomalies immediately, while honing a more calibrated detection model over the course of two weeks.
As time goes on and patterns, traffic and threats change, these solutions adapt and change as well without the need for reconfiguration.
These solutions are able to work around the clock without losing focus or needing downtime and they can be applied to as many cameras on your security network as you’d like – allowing you to put virtual eyes on every single camera in your network 24/7/365.
Systems can be set up to only present operators with the cameras that are viewing anomalies, helping to prevent operator fatigue or burnout and simultaneously improving efficiency.
In some instances, the number of operators required to monitor a given system can even be reduced, allowing management to focus their efforts on other security tasks.
As the likelihood of data breaches increases regularly and the average cost of a data breach steadily rises – up 15% from 2020 to 2023 at $4.45 million, according to IBM – private and public organizations are taking steps to secure personal identifiable data in every department.
Traditional security analytics often require storage of this type of data to reference as they attempt to detect specific types of activity within the view of a camera – transportation security is no exception.
While measures can be taken to protect this data, ultimately its existence becomes both a vulnerability and a liability no matter how secure the data is.
The only way to eliminate this vulnerability is to ensure that this data does not exist at all.
Many AI-powered anomaly detection solutions do not store any personal identifiable information, as this data is not used as a point of reference when detecting anomalies.
Anomalies are identified based solely on statistical data captured by the software in the identified surveilled area.
This unique type of anomaly detection is focused on detecting anomalous conduct and behavior, not personally identifiable information.
As a result, this anomaly detection is extremely secure and private and does not impose any added data breach liability.
Traditional video analytics solutions, while effective in forensic scenarios, often come with significant challenges related to cost and complexity.
Licensing models are often perpetual – a cost realized as a capital expenditure – and in some instances offer limited updates and bug fixes.
The resource-intense solutions also require significant hardware investments to be made, which come with ongoing costs for maintenance, updates, cooling and in many cases require a large physical footprint.
Once installed, these solutions require many hours of setup and configuration – much of which must be performed by a trained security integrator at significant cost.
One of the single largest benefits of choosing some AI-powered anomaly detection solutions is relatively low upfront and ongoing cost.
Many offerings license their products on a monthly basis, which is realized as an operating expense. With this license structure comes regular updates, bug fixes and product improvements.
What’s more, select anomaly detection platforms are extremely hardware-efficient, meaning the cost of hardware, maintenance and physical storage of the hardware is significantly lower than traditional alternatives.
Finally, configuration of many anomaly detection offerings is simple and brief – the AI begins learning and adapts entirely on its own.
As a result of these significantly reduced costs, transportation security operations which previously may not have had the available budget to add new analytics to their systems can bring safety and security improvements without lobbying for additional funding.
AI-powered anomaly detection has the power to completely revolutionize the way transportation hubs secure their facilities, while addressing some of the largest challenges faced by security teams.
With more efficient and attentive operators, consistent learning and adaptation, as well as overall low costs, this emerging anomaly detection tech can be applied quickly and help your team address incidents in real time – keeping occupants safer, faster than ever before.
This article on executive protection was originally published in the January edition of Security Journal Americas. To read your FREE digital edition, click here.