Body-worn cameras are utilized by many different workers- such as security guards, police officers, and even train conductors. They record candid videos of interactions between the camera users and the citizens around them, for the purpose of protecting both parties. But with these security measures comes some concerns- by both those using the equipment and by those captured on the footage.
Workers using body-cams may have concerns about lack of funding; The expense for the newest and most efficient technologies can be a huge hurdle, especially when better systems are needed to aid in time-sensitive cases. Body-worn cameras are a wonderful resource to help protect the public, but the time wasted from sorting through the footage is valuable. There is a need for an incredibly accurate, cost-efficient technology that sorts through video in a matter of seconds.
The public also has voiced concerns about the use of body-worn cameras and AI being used. Facial recognition systems, which save people’s faces into a database, have been said to be a privacy violation. Not only that, but Artificial Intelligence systems have a history of being biased towards specific groups. If AI is used alongside body-worn cameras, the worry about it leading to biased results is very understandable. The only solution to this is an algorithm that is not vulnerable to bias and does not identify people by use of facial recognition.
The solution? Without a doubt, it’s DejaVuAI. Here’s why:
IDing Individuals
There is an increasing number of criminals who know how to confuse AI systems with things such as adversarial noise. There are even patterns on masks and t-shirts that are used by people who know they could be on camera, but don’t want to be recognized when police review the footage. DejaVuAI is the only system currently on the market that can see through this AI/Machine Learning vulnerability. Even at different angles and bad lighting, we can show you every instance of a specific person that shows up in your footage, guiding you to leads you may have otherwise not seen, such as a building they exited from or vehicle they left in.
DejaVuAI can even ID the exact patterns designed to confuse other AI.
Prevention of Crime/Recovery Efforts
In time-sensitive cases, such as a suspicious piece of luggage dropped off, or a child going missing, our technology can be utilized to locate where these things are in footage, practically in an instant- even at different angles. We have a built-in Similarity Groups tool that can bring together related images and show you why they are related, allowing for more possible information such as another person related to the subject.
This example shows a Similarity Group with multiple degrees of separation.
Resolution is Not a Problem
Extremely low or extremely high resolution does not hinder our ability to match images. If you are following what you perceive to be a blurred, pixelated figure of a person- we can show you every instance of that blurred figure. And of course, with high resolution, we are able to do the same.
Cost-Effectiveness
Our algorithm, compared to current image recognition systems, is extremely cost-effective. A growing concern for policing is the lack of funding and being unable to obtain the newest and most efficient technologies. Luckily, we can integrate into your current systems and workflows- and we do not require new, expensive hardware. Our system is powerful, yet lightweight- as we are able to run it on a common household computer. No longer worry about the price of getting useful and powerful image recognition software.
Time-Saver
DejaVuAI is at the top of the market for image search technologies, not only because of its groundbreaking recognition abilities, but also because of the time it takes to get results. After the system is trained with whatever database you choose, finding matches usually takes literal milliseconds. Also, the time required for training is another important subject. Navigating our interface is simple, and after a short training video which explains of how to get and understand results, users are prepared to utilize the system. There is no need for someone to monitor or intervene in the actual processing of images.
Utilize our technology to quickly find your subject, even if all you have is the back of their clothes.
Unbiased AI
Artificial Intelligence systems are biased towards specific groups of people, because those who are training the initial software have their own inherent bias. Examples of this are companies who have used AI to hire employees, resulting in women being rejected because the system was biased towards male workers. Another example is a predictive policing AI, which was biased against minorities, predicting future crimes in neighborhoods with minorities despite the same amount or less crimes happening in these neighborhoods. DejaVuAI is completely unbiased, because there is no human behind the training- our system trains itself and only looks for similar patterns. As for recognizing images, it functions as a human eye would- without any purposeful or unknowing partiality.
No Facial Recognition
Also, a major concern the public has about body-worn cameras is that citizens are then put into a facial recognition system, and it is a violation of their privacy. Although our system has the power to match faces since it looks for patterns, we are not facial recognition- and we do not keep a record of people’s faces. Our system looks for other ways to identify people or locations, such as clothing, accessories (backpacks, hats, masks, etc.), tattoos or other body markings, vehicles, weapons, nearby buildings, and more. We also can be used for iris and fingerprint analysis, but this is in more applicable situations such as at a crime scene, and not a practical use for candid videos, as body-worn cameras capture.
There’s no biometric data being used (like a face wireframe), but rather we are matching the similarities in the images.
Follow us on LinkedIn and YouTube for updates. Or feel free to send us an email!
