Distance Tracking against COVID-19

The integration of advanced algorithms and models in programmable cameras.

the right distance

Covid-19 has left a major mark on the world, with impacts ranging from personal lives to industries and organizations. While the usage of Personal Protective Equipment (PPE) such as masks and gloves remains the main method to protect ourselves, the WHO has also recommended the adoption of at least one meter of distance in order to avoid the spraying of droplets.

This is leading to sudden lifestyle changes, with the adoption of barriers, physical delimiters and special paths to access shops, factories and canteens. Social group interactions are being even more disrupted, separating friends and families to avoid the spread of the virus. However, society and businesses should accept this new paradigm and find a way out to ensure productivity continues. This is even more difficult in particular industrial scenarios, where production lines have been built to maximize efficiency and continuity of production flow

How should the manufacturing industry continue to create value while at the same time ensuring employee safety? Although barriers and paths are important in suggesting correct behavior, they are not enough to ensure people’s protection, because an active monitoring system is still mandatory to prevent diffusion and track risky contacts that might happen.

How AI could help fight back

During this pandemic, we have seen many applications of artificial intelligence, mainly to identify patterns in virus spreading and to predict the contagion pace.

Considering all the recent advancements in computer vision in terms of processing algorithms and AI models, we consider them directly applicable for coronavirus monitoring: AI can help to process images coming from digital cameras, identifying peoples’ profiles and behavior, ensuring the respect of social distancing and personal protection equipment usage.

The solution

In order to have a quick and safe method to implement Covid-19 countermeasures, one of Data Reply’s approaches is based on the integration of advanced algorithms and models in programmable cameras, such as Amazon Deep Lens , which provides native cloud integration. Images are acquired by the camera and then processed locally: data should not necessarily be shared with cloud providers, totally ensuring privacy respect. This kind of camera is also cost effective, which allows a fast and economic deployment in the identified premise.

Once the algorithm reveals, according to the regulatory and premises constraints, that a distance is not correct, it can leverage the native connection to the cloud to send to specific operators a notification through the usage of cloud notifications. The supervisor of this process may receive an information about the place, and time of the violation in order to identify where and why the violation has occurred, while respecting the privacy of the operators. In addition, the model is also capable of learning the peculiarities of their equipment and specific work uniforms, such as professional caps, gowns, etc.



The scalability of the solution is guaranteed by the high security standards of the Cloud for the management of sensitive data and the security of the transmission protocol: in a complex scenario, the images are processed by cloud tools such as AWS Rekognition, with the usage of blurring capabilities in order to deal with privacy issues.



The installation of multiple cameras and the introduction of a service layer open to the interaction with legacy services such as MES, PLM and other production planning tools, helps to identify possible issues in the organization and planning of the activities in our premises (plants, canteens and other closed areas).

Combining the capabilities of modern computer vision approaches with Data Reply expertise in manufacturing scenarios, it is possible to move towards production plan optimizations (e.g. Digital Twin, predictive maintenance, etc…).

  • strip-0

    How can we help

    Data Reply is the Reply group company offering a broad range of advanced analytics and AI-powered data services. We operate across different industries and business functions, enabling them to achieve meaningful outcomes through effective use of data. We have strong competences in Big Data Engineering, Data Science and IPA; we build Big Data platforms and implement ML and AI models in a manner that is repeatable, efficient, scalable, simple and yet secure. We supports companies in combinatorial optimization processes with Quantum Computing techniques that enable an engine with high computational performances.