Gun In Hand Detection
The Gun-in-hand Detection application from Noema uses smart cameras to automatically detect the presence of weapons in restricted areas. Upon detecting a weapon, the app sends a notification or alarm, and saves video frames in which the weapon is present, for further review.
Additionally, the app runs efficiently on CPU and GPU and can be run in parallel with other apps on a single camera at the edge. Mount a new camera or equip the application to an existing one with no additional hardware required.
Gun-in-hand detection application uses human behavior tracking to vastly increase the accuracy of detection, specifically when weapon is brandished but not perfectly visible. The application also reduces instances of false alarms.
When it comes to shooting incidents, every second matters. Even where cameras and operators are present, it often takes too long to dispatch law enforcement and first responders to scenes of violence.
Camera systems with human operators currently lack the monitoring bandwidth, consistency, and accuracy to alert first responders in a timely manner.
Noema’s Gun-in-hand Detection app uses smart cameras and computer vision to continuously monitors specific areas for the presence of firearms and other restricted items. The app automatically saves and timestamps video frames where a weapon is detected. It was designed to quickly alert for instances of firearms, more effectively and reliably than a human operator.
Directly detect the gun as well as restricted items
Runs at the edge (CV algorithms performed on the smart camera)
30 FPS, 95 Detection Rate
Modified live output
High inference speed
Many camera types are supported
Easy configuration & use
The app was trained extensively with regards to human behavior, making the app much smarter. By monitoring behavior, the app looks for instances specifically where a firearm is being brandished (in someone’s hand), rather than simply looking for instances where a firearm is present in the frame. This makes the app robust against situations where the gun isn’t perfectly visible to the camera, but the person is still brandishing it.
The app was trained with fisheye cameras, making it compatible with fisheye setups and useful for monitoring a wide area with a single camera.
The app runs efficiently on the camera hardware. making it possible to combine gun-in-hand detection with other edge applications in Noema’s portfolio, such as Seat Occupancy Monitoring.
The App translates Metadata to ONVIF XML Schema
Supports Message Broker
Supports Data Trolley
The app generates metadata enabling 3rd party system integrations.
Azena OS 1.4.0 or Compatible
Android 8.1 or Compatible
Supports a variety of camera types and manufacturers, including CCTV cameras
The application can be applied to cameras that physically or electronically pan, tilt, and zoom (PTZ)
Remote installation and configuration with intuitive interface to define monitoring areas
On-demand integration with custom MessageBroker and DataTrolley information system performed by Noema.
App is running on edge device
App is running on edge device
VMS show the camera footage along with some added featurese
SCADA handles the metadata that is generated by the app
Video pixels will show fire & smoke data on front end.
Data integrated into dashboard enabling real-time analysis