For many years, video analytics have enabled end users to detect specific people or vehicles entering restricted areas, capture license plate information, scan crowds for specific individuals and much more. Today’s Video Management Systems (VMS) and IP cameras are built with powerful processing capabilities, helping to drive the development of more advanced analytics to more accurately detect abnormalities in behaviors that trigger alerts – an important component of predictive analysis.
What Is Predictive Analysis?
With this greater number and variety of data points available to security professionals, the trend toward integrated solutions is fueling growth in the evolving science of predictive analysis. Differing from standard alert triggers, predictive analysis uses information gathered from a wide range of data sources including surveillance, access control, visitor management, incident management and other systems, evaluating the information against established behavioral models, taking into account, earlier incidents, effectively predicting the likelihood of a similar event in the future.
Note that predictive analysis does not always generate binary security events, but typically identifies occurrences such as irregular traffic patterns, motion detection during off peak hours or in restricted areas, and correlates this data with facial detection, access control activity or other system information to indicate anomalous activity or the potential for an incident of interest to occur. When this happens, the system can alert security professionals to take proactive preventive actions.
Predictive Analytics In IP Cameras
Camera-based analytics have improved significantly in recent years but these technologies remain largely reactive, providing valuable information for post-event investigation and follow-up. However, as the security industry continues its momentum toward a more predictive model, intelligent IP cameras will play a key role in allowing security staff to take more meaningful, proactive actions to prevent incidents before they can occur.
Those cameras equipped with video analytics are already driving advancements in predictive analysis, by providing the means to more accurately detect incidents and events and turn those into inputs which provide context systems can be designed to indicate compromising activity, enabling a fast and highly informed response, potentially highlighting potential future risks before they can occur.
People-counting video analytics can be used to detect a large number of people gathering in a particular area off hours
The below real-world example illustrates the value of intelligent IP cameras in enhancing predictive analysis to improve security personnel’s ability to take proactive rather than reactive action and avert threats.
Detecting Unusual Behavior
For example, people-counting video analytics could be used to detect a large number of people gathering in a particular area during off hours. This could be as benign as people gathering to celebrate a coworker’s birthday or it could be something much more sinister, such as a group of disgruntled employees coordinating the theft of company assets and/or data. Without context, however, it’s impossible to tell where on the potential threat spectrum this event would fall.
As the predictive solution begins gathering contextual data for analysis, video analytics can determine whether an alert should be issued. Based on predetermined factors or analysis of prior events, such an off-hours gathering may be enough for the system to alert security staff to a potentially negative situation. The group’s location could also be a factor, as a gathering in a conference or break room would be much lower priority than if it were in a restricted area.
Alerts can also automatically trigger cameras to switch to higher resolution, initiate both on-board and remote recording, maintain focus on a particular area and/or entry point, and even launch facial recognition analytics. This is helpful for security management and staff when responding to an alert, providing more complete understanding of the situation to determine the appropriate course of action.
|Access control data combined with people-counting may indicate that the number of people assembled exceeds the number of card swipes|
Alerts also initiate contextual analysis to determine the specifics of a situation. Technology can’t accurately judge intent; it collects all available information to determine the level of severity of a situation and, in turn, what actions need to be taken – both in the immediate term and post-event.
In the case of a large after-hours gathering, data collected and correlated from additional systems and sources will provide fuller context. For example, access control data combined with people-counting may indicate that the number of people assembled exceeds the number of card swipes. While multiple people entering on a single card swipe is a clear violation of policies, it still may not indicate a threat. Several authorized individuals may have arrived at the door at the same time and simply entered together out of convenience. Using facial recognition both at the edge and within the VMS to compare those who entered against the database of authorized users will determine whether there may be a potential security threat from unauthorized individuals being in a restricted area.
Once data has been gathered to provide greater context, predictive solutions analyse all available information to determine the level of the threat. At the same time, the system correlates information from intelligent cameras, access control, incident management and other sources to build a profile of either normal or abnormal behavior that will be used to analyze similar occurrences in the future that may or may not indicate a potential compromise.
Once data has been gathered to provide greater context, predictive solutions analyze all available information to determine the level of threat
Information from network-based calendar solutions may reveal that the gathering was simply a scheduled meeting or training. People-counting and facial recognition analytics combined with access control and HR system data could show that all of the individuals present are authorized to be in that particular area and that the discrepancy between the number of card swipes and individuals is indeed the result of tailgating. If this is the case, an email could be sent to employees to remind them of security policies.
However, if unauthorized individuals are found to be in the area, security staff would likely be dispatched to that location to determine the purpose of the gathering and ensure that unauthorized users are removed from the area in compliance with established policies. At the same time, the physical and IT access credentials of those in attendance can be temporarily deactivated to reduce the risk of insider theft. Additionally, intelligent cameras can employ analytics to detect whether objects have been removed from the area and isolate video of any such incidents for response.
As illustrated by the above example, intelligent video cameras play a significant role in more effective predictive analysis. Integrated security and surveillance systems with powerful video analytics can be deployed to improve security, lower risk, reduce fraud and transform traditional security operations from a simply reactive to much more proactive function. The intelligence gleaned from each incident empowers security professionals with the opportunity to avoid potentially dangerous situations before they can even occur.