Laptop Ban, KL Airport Attack Highlight need to Overhaul Checkpoint Security Globally
The recent ban by US and UK authorities on laptops and other computing devices being brought on-board airplanes and the assassination of the North Korean leader’s half-brother at Malaysia’s main airport have highlighted the need for a review of surveillance at major international checkpoints.
In recent years, low-cost carriers have contributed to a surge in international air travel. Major land infrastructure projects such as the US$11 billion Zhuhai-Macau-Hong Kong bridge and the just-approved High Speed Rail connecting Singapore and several Malaysian cities are also posing serious challenges to security officials in charge of border crossings.
As land, sea and air travel has increased many times over the last few decades, much of the security focus has been on advances in body and baggage scanners, especially post 9/11; increasing the numbers of surveillance cameras; or making the resolution of such cameras sharper or more “intelligent”.
But criminals and terrorists have also gotten smarter. The female attackers of Kim Jong-Nam, half-brother of North Korean leader Kim Jong-Un, used a toxic VX nerve agent – a liquid which could easily avoid detection – at Kuala Lumpur International Airport despite hundreds of CCTV (closed-circuit television) cameras.
Surveillance remains at the heart of almost all solutions to prevent security breaches, but it is not the number of cameras or their sophistication that counts.
Amid forecasts of the continued growth not just of air travel but the number of airports themselves, as well as of many cross-border land-transport hubs, two key attributes are necessary in this age of heightened threats of criminal and terrorist activities.
First, beyond camera resolution, the entire surveillance approach needs to be driven by artificial intelligence (AI) that can “intuit” or predict the potential of wrongdoing based on advanced algorithms that assess apparent scenarios based on images, sounds or video footage.
Using AI, Graymatics has worked with security agencies to “teach” their surveillance systems to look for unique insights and hidden relationships between data sources – without being explicitly programmed on where or when to find them.
For example, many large enterprises in South-east Asian countries have used such a platform to monitor and maintain security by detecting intruders in a restricted area, recognizing suspicious activities, detecting violation of traffic laws, tracking vehicles through number plate detection and many more.
The second is the need for highly intelligent tools to conduct search of CCTV archival footage easily and rapidly during and immediately after an emergency. In the aftermath of an attack or a breach, security officials will need to go through thousands of hours of video footage to look for patterns or even collaborators involved, days, if not weeks, before the event.
For example, image and video analytics can detect any suspicious or unethical activity in the vicinity which a security officer, viewing images on a bank of screens on his or her own, cannot. The key lies in deploying AI and algorithms to cut down on man-hours required to sift through footage to find the perpetrators or the patterns they deployed.
This use of AI is especially significant when we consider that, for the most part, the science of analytics has been limited to binary and textual (alphanumeric) data. Big data for image, audio and video is far more complex.
By applying an image-focused subset of machine learning, specific visual details and metadata – such as a “blue cap” or a “yellow fedora” – can be found among many months of archival footage in as quickly as just a few hours. Image recognition technology can even employ facial recognition, vehicle classification and object tracking in real-time as they rapidly scan CCTV archives.
Over time, not only can the surveillance system be equipped to search and intuit human observation, but cameras can even learn to pan towards and zoom in on situations that are potentially threatening. Different cameras can work as “teams” to analyze and counter specific threats, and microphones can be trained or programmed to detect tones of voices or sounds normally associated with a threat.
Such technology is no longer science fiction. Well-installed, a user-friendly dashboard able to analyze multiple cameras feeds in real time can enable a few security personnel to predict or respond quickly to threats or concerns. Such a tool can save precious minutes and can well avert a major crisis.