Milestone Systems partners NVIDIA to launch Project Hafnia for AI model training

Milestone Systems has announced the launch of Project Hafnia, a powerful new platform designed to give AI developers access to high-quality video data for training visual AI models. This initiative is being rolled out in collaboration with NVIDIA, using its Cosmos Curator and AI fine-tuning microservices to deliver cutting-edge data curation and training capabilities.

What is Project Hafnia?

Project Hafnia is Milestone’s ambitious plan to democratize access to video data for AI training, starting with visual models used in traffic and transportation systems. It aims to help developers overcome a major challenge, the lack of reliable and compliant video data needed to train AI models effectively.

By using NVIDIA’s advanced tools like the VSS (Video Search and Summarization) AI blueprint and fine-tuning services, the platform delivers enhanced speed and accuracy for model development, particularly optimized for NVIDIA GPUs.

Who Is It For?

The platform is meant for two major audiences:

1. Data generators, such as Milestone’s global partners and customers, who want to share and use their video data securely and compliantly.
2. AI developers, who need access to curated, annotated video data to train their models — especially in areas like traffic management, smart cities, and law enforcement.

Why It Matters

“Access to high-quality training data is one of the biggest bottlenecks in AI development,” said Thomas Jensen, CEO of Milestone Systems. “With Project Hafnia, we aim to build the world’s smartest, fastest, and most responsible platform for AI model training.”

According to Milestone, using Project Hafnia could speed up AI development by up to 30 times, thanks to its automated and scalable data curation capabilities powered by NVIDIA.

What’s Next?

The first offering under Project Hafnia is a Visual Language Model (VLM) for traffic analytics, trained using real transportation data and fine-tuned on NVIDIA’s infrastructure. It can be used for tasks like traffic condition assessment, incident reporting, and alert validation.

Project Hafnia is currently in its pilot phase, and developers can sign up for early access at their official website.