Milestone is one of the early adopters of the newly announced NVIDIA Omniverse Blueprint for Smart City AI, a reference framework for optimizing city operations with digital twins and AI agents. Milestone is also expanding its data platform with NVIDIA Cosmos to generate synthetic video data from real world inputs. Leveraging both real and synthetic data, Milestone will build and train vision language models (VLMs) responsibly. European cloud provider Nebius will provide the required GPU compute to train the models.
AI opens up new horizons for smarter, more efficient cities with Project Hafnia’s data platform – from intelligent traffic and transportation management to better safety and security for people and property.
VLMs learn to map the relationships between text data and visual data such as images or videos, allowing these AI-models to generate summaries and insights from visual inputs. Milestone Systems is working with NVIDIA to empower European cities like Genoa to build and fine-tune computer vision and AI applications on a foundation of fully compliant and ethically sourced data. The project is rooted in regulatory integrity, data diversity, and AI relevance, aligning with the EU’s legal frameworks, including GDPR and the AI Act. This ensures both transparency and fairness in the development of AI technologies.
“I’m proud that with Project Hafnia we are introducing the world’s first platform to meet the EU’s regulatory standards, powered by NVIDIA technology. With Nebius as our European cloud provider, we can now enable compliant, high-quality video data for training vision AI models — fully anchored in Europe. This marks an important step forward in supporting the EU’s commitment to transparency, fairness, and regulatory oversight in AI and technology — the foundation for responsible AI innovation,” says Thomas Jensen, CEO of Milestone.
Milestone’s Project Hafnia provides the missing link for developing video-centric AI innovation with a fully compliant and ethically sourced data library for fine tuning classical video analytics models and VLMs.
The VLM’s accuracy and performance optimizations are tuned for running on NVIDIA GPUs and in NVIDIA AI Blueprint for video search and summarization (VSS).
