Frequently Asked Questions
General
If a true weapon is detected, an alert is sent to ZeroEyes' in-house security monitoring team. Once verified, it can be sent to a local emergency dispatch (such as a 911 call center), onsite security staff, police and school administrators. We have dozens of options for 3rd party integrations so you can see the alert in the system you use today. At the same time, our analysts are trained to contact your security team directly via phone to a call tree. This process is as fast as three to five seconds and accelerates the traditional dispatch process.
The most important information we provide is the exact location of the camera, the exact time of the incident and the image of the detection with a bounding box identifying the threat object. This provides first responders with a description of the individual, the weapon system they are carrying and where they are located, reducing confusion and in many cases, improving response time.
We strive to deliver alerts in as near real-time as possible. Our AI processes real-time video in milliseconds and we can dispatch alerts in as fast as 3 - 5 seconds. Client site networks are often the limiting factor and without strong wifi or mobile cellular connection, delays may occur.
We built our weapon detection capability with hundreds of different types of weapons, particularly on those most used in mass shootings, such as pistols and AR-style rifles. We are constantly improving and adding to our AI model.
No, we do not detect concealed or holstered weapons. The weapon must be visible to security cameras and clearly identifiable for our AI to detect.
Detection distance is determined by image quality factors such as resolution, pixel density, and zoom configuration. A general rule of thumb is the object must be at least 0.1% of pixels in the video frame with adequate lighting to be detected.
As fast as 3 - 5 seconds. We have over 1,000 verified dispatches of actual weapons at client sites.
Our in-house ZeroEyes Operations Center works 24/7/365. We are alerted of any potential threats at any time of day or night, and dispatch these to our customers and law enforcement if needed.
Technology
ZeroEyes is deployed on-premise and configured to your environment and requirements.
When we founded ZeroEyes with the mission to end gun violence, we discovered quickly that human-ai teaming offered the best performance for identifying threats. We are AI-powered and human-driven. We optimize for maximizing detection confidence before minimizing false positives because our expert analysts filter false positives and only dispatch human-verified alerts. False positives vary greatly depending on human activity in camera view, but we take care of the false positives so you can trust when you receive an alert from ZeroEyes, it is a real threat that requires response.
Privacy
No, we do not use or offer facial recognition technology or any form of biometric identification. Our AI is trained only to look for visible threat objects like weapons. The only time a person's face is visible in alert images or video is when a threat object has been detected and there is suspicion of a threat to human life.
Our technology sits on top of an organization's current IP security cameras, which could also have facial recognition software installed by the organization. ZeroEyes does not manage or interact with that software.
If allowed by the customer, we record video of our technicians onsite in the process of testing detections during the installation phase. We store detection images for improving our AI and providing forensic detection reports. This data is always transmitted and stored in an encrypted format and data retention timeframes can be specified by the customer.
No, for the privacy of all, we do not store any biometric identifiers. Biometric identifiers are unique, measurable physical or behavioral characteristics used to automatically verify a person's identity. Common examples include fingerprints, facial patterns, iris/retina scans, and voice, which are converted into digital data.