Irisity AB (publ) is launching the deep-learning algorithm IRIS Violent Behavior for increased safety in public places.

Real-time detection of potentially harmful behavior in public and private areas is a sought-after but challenging task for surveillance systems and security providers. So far, this detection has mostly been done manually at high cost and low precision.

Irisity’s new deep-learning algorithm IRIS Violent Behavior analyzes with high precision and in real time an unlimited number of cameras and triggers video alarms when violent behavior is identified. The algorithm uses pose estimation technology to determine the exact position of selected body parts and has been trained on millions of images. IRIS Violent Behavior identifies situations with violent or otherwise unusual human motion, including situations likely caused by violence, such as people sitting or lying down on the ground.

The service can be easily installed on both existing and new cameras and thereafter protects public places around the clock.

Irisity’s new algorithm is included as an add-on to the SaaS service. IRIS is fully scalable for large installations and can be selected together with Irisity’s patented real-time anonymization for the protection of personal privacy.

IRIS Violent Behavior is now offered to customers in Sweden and internationally together with our global partner G4S.