Innovation in AI and two-step verification is reducing the labor to monitor cameras - critical as surveillance continues to proliferate.
The longstanding practice of watching hundreds and thousands of cameras for suspicious behavior – and then reacting – is over. This method has proven ineffective, especially as surveillance continues to proliferate in home, business, smart cities and other connected environments. In addition, conventional live-video monitoring services tend to not mention the total inherent delay times from the detection of an intrusion to the execution of effective deterrence reactions.
With the ongoing shortage of labor and contract guard services, as well as humans who simply can’t stay attentive to multiple video displays, technology is stepping up to assist and revolutionize remote monitoring services.
Sensing, detection, analytics and artificial intelligence (AI) have changed the formula of monitoring from an after-the-fact forensic activity to a proactive and strategic tool that can actually deter and prevent crime and property loss, at a lower total cost of ownership to the user. Now, the emphasis is on automation of video monitoring by pinpointing the alarms that really matter, significantly reducing the labor and customer cost to monitor cameras and ultimately signaling a step up in the demand for video alarm monitoring services.
All eyes on video
The use of security cameras and human surveillance to protect people and property has exploded over the last few decades. Interestingly, property crimes in the U.S. have dropped from 4,740 per 100,000 people in 1991 to 2,110 per 100,000 people in 2019. It is likely the adoption of security cameras and surveillance has influenced this positive trend. Also, according to a recent report by PEW Research Center, “just one-third (34%) of all property crimes are reported, with the general feeling being that the police would not or could not do anything to help. Others felt that the crimes were either too trivial or more of a personal nature.”
According to FBI crime data, in 2020 stolen property loss in the U.S. was estimated at $971 billion versus recovered value of $54 billion for a recovery rate of 5.6%. The burglary trend in 2022 is heading upward as evidenced by reports from the New York Police Department (NYPD) and Chicago Police Department (CPD) where burglaries have increased 32.7% and 35%, respectively in same period comparisons of 2022 to 2021.
The key takeaways from the crime statistics are the reappearance of an upward trend in burglary and an extremely low loss recovery rate. Cameras as a post-crime forensic tool with only a 5.6% burglary recovery rate has a poor payback. Video monitoring with 911 dispatch capability is both expensive and dependent upon the response time of authorities. Technology is the answer. Where do we go from here? The answer is in the application of technology.
There is significant potential in using deep-learning artificial intelligence to not only automate deterrence of intrusion but also to automate a significant proportion of the human labor required to monitor video. Crime occurs in seconds. Time matters! Automating deterrence with randomized and unpredictable actions offers an immediate response and is proven effective in thwarting unwanted intrusion. Further, adoption of deep-learning technology to automate video monitoring offers a reduction in labor cost which enables more attractive pricing for this RMR service and ultimately an increase in the demand for video surveillance services.
The state of RVM
Interviews with remote video monitoring (RVM) businesses has yielded a general understanding of the current operational state and overall cost picture. Most RVM business models require monitoring personnel to verify the validity of all events detected on a client’s property. This is followed by human assessment of the intent of the intrusion (e.g., casual intruder, loiterer, criminal intention, employee, cleaning crew, security guard, etc.). If the intruder exhibits ill intent or persistently loiters, monitoring personnel can engage the intruder with voice down messages and alarms and, if necessary, dispatch authorities. Some RVM providers may also schedule automated deterrence actions while continuing to monitor the reaction of the intruder. To reduce false positives, the industry is quickly adopting and testing various cloud or edge video analytics and AI tools. Typical monitoring productivity seems to average between 150 to 200 cameras/person resulting in burdened labor costs around $20 to $25/camera/month.
Development teams are training convolutional neural networks (CNN) algorithms within the AI application to accurately and quickly detect both intrusion and persistent loitering for use in fully automating proportional deterrence reactions. In deep learning, CNN is a class of artificial neural network (ANN), most commonly applied to analyze visual imagery.
Upon review of tens of thousands of recorded intrusions on commercial properties, we have proven that 99% of unwanted intrusions were successfully deterred without the need for human interaction. The characteristics of these intrusions and deterrence events have been extremely useful in shaping our understanding of when it is important to engage human intervention and of course when it is needless. Conservative estimates have convinced us that an 80% reduction in the labor cost to monitor cameras is achievable when we properly train CNN technology to automate a significant portion of the monitoring task. This represents a reduction in labor cost from $25 to $5/camera/month. The efficiency is realized from both the classification technology combined with an operator interface that easily communicates event priorities.
Deep-learning artificial intelligence is the future of video monitoring. Proper application successfully automates deterrence of unwanted intrusions and focuses human operators on the real threats and safety concerns that require intervention and possibly 911 dispatch. The result of this technology evolution is lower costs for both the service provider and property owner - driving higher market adoption and continued declines in burglary trends.
Greg Ayres is the Vice President of Marketing and Business Development for iDter (www.idter.com), a San Mateo, Calif., -based technology company focused on proactive deterrence of criminal activity and protection of open-air assets. He previously was a GM with Honeywell Industrial Controls Division and a Chemical Engineer with the Procter & Gamble Company.
This article appeared in Security Today, Sept./Oct. 2022 issue