IRS22 is based on the same technology of deep learning and image recognition that is tested and highly scalable, developed by a team of researchers from the University of Washington in collaboration with the Allen Institute for Artificial Intelligence and Facebook AI Research</a > which allows to analyze billions of daily images and videos.
The ErInformatica team has extended and engineered the image recognition technology in a product suitable for real-time analysis of video flows in heterogeneous contexts.
- Large library of objects already trained for recognition
- Possibility to train the system to recognize categories of objects not included in the initial preset
- Real time processing capacity
- Automatic generation
Through the image recognition technology, the system can identify objects not allowed (in specific analysis areas) triggering the generation of a real-time alarm, which can be sent to third-party systems.
The system is able to analyze the presence of one or more types of objects in a specific area. This analysis can be historicized for statistical purposes.
Users analysis in ski area
- User (skier or snowboarder) in a non-admitted area (let us suppose off-piste)
- User (skier or snowboarder) on the track beyond the guaranteed time (suppose the evening time, the alarm can be propagated to the snowcat drivers in order to prevent a possible collision)
- Motorized vehicle on the track during track opening hours (let us suppose a snowmobile that intends to drive one the track during opening hours)
- User with unsuitable equipment (we assume bob or sled) on the track
Analysis of industrial environments
- Heavy vehicle unloading / loading areas
- Analysis of lanes dedicated to forklifts
- Analysis of areas restricted to users during specific processing (Steelworks, Foundries, Chemical plants)