crowdHEDGE: using AI to keep citizens safe


How Addfor leverages a Lenovo ThinkSystem SE350 server, powered by NVIDIA® T4 Tensor Core GPUs, to collect, process, and analyze drone video footage at the edge to help strengthen COVID-19 public health measures.

“The Drones Unit of the City of Turin has successfully tested, within the services of the Public Administration, the use of drones, AI, and edge computing solutions with the collaboration of Lenovo, NVIDIA, and Addfor. The use cases developed are related to the real-time analysis of data, which are useful for developing innovative services for citizens. Thanks to this pilot project in the field of safety and security of public spaces, the City of Turin addresses the combined use of drones, AI, and edge computing to focus on urban and aerospace mobility as an opportunity for the use of AI within the Public Administration”

Marco Pironti

Deputy Mayor Innovation and Smart City
City of Turin.

CrowdHEDGE is an Artificial Intelligence that extracts data insights from drones’ video streams flying on public spaces, such as city parks and recreational areas. Addfor teamed up with the City of Turin to connect crowdHEDGE to drone video streams. The aim was to enable public health officials to track metrics such as foot traffic and crowd sizes and use an advanced graphical interface to explore the data by time and geographical area.
Even as large-scale immunization initiatives get underway, social distancing, face coverings, and contact tracing remain some of the essential tools that public health bodies have to reduce coronavirus spread.

Like many European countries, Italy put social distancing measures in general settings such as shops, offices, public transport, and public outdoor spaces.
However, the effectiveness of these measures depends on people carefully following the guidelines.
Enrico Busto, Founding Partner & CTO at Addfor, explains:

Measuring compliance with COVID-19 public health measures poses a huge challenge for authorities. We set out to develop an AI-enabled image recognition system that could be used in the fight against coronavirus.”

How could the company process vast amounts of data rapidly and cost-effectively on behalf of the City of Turin?

To support the crowdHEDGE pilot in Turin, Addfor, an NVIDIA Metropolis partner, implemented a Lenovo ThinkSystem SE350 edge server, equipped with NVIDIA® T4 Tensor Core GPUs.
The Lenovo ThinkSystem SE350 is a purpose-built, ruggedized, highly secure edge server specifically designed to support the compute, storage, and connectivity requirements at the edge. The Lenovo ThinkSystem SE350 has operating temperature ranges from 0-55°C and is targeted for deployment outside of a data center environment using wall or ceiling mounting systems. It can be deployed, managed, and repaired with nZTP (near zero touch) almost anywhere, even in a van or mobile environment. It is compliant with NEBS3 and ETSI standards, thereby passing rigorous emission, altitude, shock, and vibration testing. Built-in WiFi connectivity enables Addfor to wirelessly stream video data directly from drone video streams at the edge where the video is captured, maintaining privacy.

“Processing and analyzing data at the edge reduces traffic load and improves performance,” says Enrico Busto. “The Lenovo ThinkSystem SE350 is, therefore, the ideal solution for this project. It’s a small, rugged, ultra- reliable server that enables us to collect, process, and analyze vast amounts of video data rapidly, close to the source.”

Furthermore, the Lenovo ThinkSystem SE350 is the most compact server delivering NVIDIA T4 GPUs, making it a performance leader for edge AI workloads that demand the highest performance.

“We were particularly impressed with the support for NVIDIA T4 GPUs, which made the Lenovo ThinkSystem SE350 perfectly suited to our needs,” comments Enrico Busto. “For example, GPU-acceleration for analytics workloads enables us to calculate the number of people present in a given area in real-time.”


Backed by Lenovo and NVIDIA technology, Addfor helps the authorities in Turin monitor compliance with COVID-19 related public health policies. The crowdHEDGE solution can calculate the approximate distance between individuals and identify the frequency of risky social contacts based on each gathering’s size and the amount of time spent together.
Enrico Busto comments: “We anticipate that crowdHEDGE will play an important role in helping local authorities track the spread of coronavirus. This insight will empower them to make data-driven decisions to protect public health.” While the City of Turin is currently using crowdHEDGE to help in the fight against COVID19, the solution has a wide range of other potential applications.

“We’re proud of the achievements that we have made with crowdHEDGE,” says Enrico Busto. “We are already exploring ways that can adapt the solution to prevent overcrowding in business and public settings, or even help police forces better protect citizens.”