COVID-19 global pandemic has transformed priorities around the world, creating the need to measure human interactions to ensure effective prevention of infection during the called New Normal.
Social distance, therefore, is one of the most effective measures to prevent the transmission of SARS-CoV-2 and will remain in place until the virus is defeated.
The new problem to be solved
This paradigm gives rise to a new problem involving:
- How to calculate the social distance with monocular images regardless of perspective.
- How to use a single model of artificial vision that does not have to be calibrated for each camera.
- How to optimize video stream processing on the edge, ideally in the camera itself.
“neuroons Social Distancing”, the solution for the New Normal
Given this situation, neuroons’ team identifies the ubiquity of IP cameras to apply artificial vision models. Our solution offers:
- Alerts and integration with warning systems, ensuring both regulatory compliance and privacy, as the software does not store video, images, or personal data.
- A monitoring panel, since the machine vision model is pre-trained or retrainable for particular scenes.
- A machine vision model optimized for edge processing.
neuroons Social Distance can be applied to flow, capacity and distance measurement in public environments: shopping centres, schools, hospitals, beaches, public transport, cultural and sports venues.
Our approach to responding to this impending need is based on a neural network architecture that predicts 3D locations and corresponding confidence intervals from 2D human poses.