Why Computer Vision is Essential to the Future of Corrections
Ken Dalley
On October 12, 2017, four inmates attempted the deadliest prison escape in North Carolina history. Mikel Brady, the inmate ringleader, conceived a plot to keep checked-out tools accessible to inmate workers in the prison industries’ sewing plant at Pasquotank Correctional Institution. Six-inch blade scissors, two screwdrivers, a ball-peen hammer, and a claw hammer, were distributed to three other inmates the day of the attempted escape and used as weapons of convenience to bludgeon and hack their way to the prison’s perimeter fencing, resulting in the murders of four prison employees.
At trial, Brady admitted that he had been plotting the escape for months. He knew which doors were unlocked, and where blind spots existed to hide from security cameras. Brady knew that staffing was short, making it easier to fabricate spears out of supplies in the warehouse using a wooden broom and metal from shelving. And Brady knew that allowing inmates to distribute tools to each other was a common violation of security protocols.
Dangers often lurk in plain sight. That’s why computer vision is essential to the future of corrections and the safety of correctional officers and inmates alike.
There’s no question that security camera systems have an important role to play in the surveillance and operational security of a correctional facility. But too often, security camera systems are used in a limited fashion as an investigative tool when an altercation or allegation arises.
As the expression goes: if a picture is worth a thousand words, a video is worth a million. And computer vision has the unique capability to unlock exponential value when paired with either.
Computer vision is a specialized field of artificial intelligence that aims to extract useful information from videos and images, such as people and objects. Tagging our family and friends in the photos we take with our smartphones is a common use of computer vision. Computer vision is used for medical image analysis to detect disease, such as skin and breast cancer. Cars use pedestrian detection technology, combining sensors and computer vision that integrate with our vehicle’s braking systems to keep safe distances from cyclists and dog walkers.
Our world is surrounded by valuable, everyday uses of computer vision and has been for several years.
The stark reality is that correctional officers face one of the highest rates of non-fatal, work-related injuries of any profession in the United States. The rate of injury is higher in corrections than for firefighters and police officers. Many injuries to correctional officers are sustained by weapons fashioned out of ordinary items, made, or received by inmates in unauthorized locations or places with little to no supervision.
With a high rate of workplace injury, combined with 60% of corrections officers attributing their workplace stress to staffing levels, workplace policies, and poor technology, according to a One Voice United report, the need to modernize the correctional officer has never been more important.
Where and How Can Computer Vision Help Correctional Officers?
In the Pasquotank prison break, the U.S. Department of Justice investigation discovered inmates were free to roam unobserved near the prison-industries sewing plant. This location is where several inmates started a fire to create a diversion during their breakout attempt. Well-positioned security cameras can deter misconduct but can’t outright stop it. However, by using facial recognition as a layering technology, you can biometrically identify those in the field of view, cross-referencing identities against a list of approved inmate workers. This is particularly useful for distinguishing between those who are expected to be present during shift work, versus those not on shift. And in the case of any unauthorized presence, the technology immediately notifies the staff of the discrepancy.
Jails often have this problem, too. Without proper supervision, inmate worker areas can attract non-inmate workers, resulting in unsanctioned loitering and fraternizing that can escalate to contraband exchanges as well as physical and sexual assaults.
Facial recognition offers three distinct advantages:
1. It’s near perfect in accuracy, particularly in environments where camera placement is optimal, and lighting is sufficiently illuminated.
3. It’s contactless, where other forms of identification—whether electronic or biometric—often require contact or close-range contact.
3. Its ease of integration with existing video management systems, while built to easily identify one or many individuals simultaneously.
A common use case for facial recognition is inmate identification or ingress or egress to verify inmates are in the right place at the right time. But facial recognition is also ideal to ensure that the correct inmate is being released from custody, particularly when combined with multi-factor authentication (MFA), such as the inmate’s RFID card or wristband. Inmates may provide a false identity, or they may resemble another inmate. When staff fail to follow release protocols, the wrong inmate gets released and law enforcement exhausts time and resources to recapture the missing offender, potentially placing communities at risk.
The security benefits of computer vision aren’t limited to just correctional officers. It provides an equally important safeguard for inmates as well. The Commissions on Safety and Abuse in America’s Prisons and the National Prison Rape Elimination Commission found that violence in prisons deters inmates from reintegrating into society successfully. In a Department of Justice report, 20% of male inmates reported experiencing some form of physical violence by other inmates.
Corrections expert Gary Cornelius has detailed various reasons why inmates are violent and the correct countermeasures and controls to use to prevent this violence. Improvements can start immediately during the booking, intake, and classification process. Stronger observation skills among all correctional officers are crucial to security and curbing violence as Cornelius explains, “Assaultive inmates know which staff are not attentive to security, which staff sit too long at their posts, and where the blind spots are in a facility.”
Mikel Brady is a textbook demonstration of an inmate exploiting the vulnerabilities of a detention facility. Brady observed the security lapses in the sewing plant area for months. He knew staffing levels were low, and that protocols were not being faithfully followed. Brady knew where staff would be blind to certain camera views.
Computer vision enables common objects to be trained and identified as easily as human visual systems. Every common object that is accessible to inmate workers, such as blade scissors, screwdrivers, a ball-peen hammer, and a claw hammer hung on a shadow wall, can be trained, and easily seen by computer vision systems. When combined with facial recognition, an identified person can be seen in computer vision taking a claw hammer at a specific time, and a configurable notification can be generated if the item is not returned at a certain time, or at all.
Computer vision is effectively a “checks and balances” technology for correctional officers. The purpose of computer vision is to never replace staff, but to augment their awareness.
Proper staffing levels in jails and prisons will likely remain a forever struggle. However, the shallow use of security cameras and video management systems is one of the most overlooked and critically underdeveloped capabilities that is as much a security necessity as it is a moral one. Computer vision has the capability to see what correctional officers miss and inform them of risks and threats that officers may be blind to. In fact, computer vision democratizes security—providing equal protection to correctional officers and inmates alike.
In 1984, the United States Supreme Court held that inmates have no reasonable expectation of privacy in their cells under the Fourth and Fourteenth Amendments. In one of the most surveilled industries the world has known, in which most facilities have blanket coverage of video cameras in a jail or prison, there’s a virtual innate preparedness to bring computer vision into the fold.
To create real, systemic change that can protect correctional officers, inmates, and every stakeholder in corrections, we have the responsibility to better align public policy and legislation. Computer vision and facial recognition technology initiatives can enjoy consensus and bipartisan support. Right now, various forms of legislation are being drafted to protect personal freedoms from unauthorized biometric generation and surveillance. Other states, such as Texas, Illinois, California, Arkansas, Florida, and others have enacted privacy acts, such as the Capture or Use of Biometric Identifiers (CUBI) Act, in which organizations are generally prohibited from capturing biometric identifiers for a commercial purpose unless they first provide notice and obtain consent from the affected individual.
Restrictions or prohibitions from computer vision and facial recognition in the general public are important, particularly if the risks to personal privacy outweigh its benefits. However, in specialized environments such as jails and prisons, where staffing levels and risks to bodily injury will remain a forever struggle, and the U.S. Supreme Court has held that inmates have no reasonable expectation of privacy in their cells under the Fourth and Fourteenth Amendments, computer vision is an essential technology not just for every correctional officer, but for the future of corrections.
______________________________________________
Ken Dalley is the founder, chairman, and Chief Warrior of GUARDIAN RFID. He holds several patents in RFID, artificial intelligence, mobile, and Cloud computing. Ken received a Bachelor of Arts in English from St. John University. For more information, he can be reached at ken@guardianrfid.com.
References
Dawe, B., Potter, A., Greer, S., Post, C., Carrillo-Klein, H., Tripp, J. Torretta, O. (2021). I AM NOT OKAY. One Voice United. https://onevoiceunited.org/wp-content/uploads/2021/10/Wellness_White_Paper_Edited_OCT17_2021.pdf
Wolff, N., Blitz, C. L., Shi, J., Siegel, J., & Bachman, R. (2007). Physical Violence Inside Prisons: Rates of Victimization. U.S. Department of Justice: Office of Justice Programs. https://www.ojp.gov/ncjrs/virtual-library/abstracts/physical-violence-inside-prisons-rates-victimization