Milestone Systems enhances AI model training with seamless video data transfer

Milestone Systems has launched Project Hafnia, an initiative to enhance AI model training through seamless, compliant video data transfer using NVIDIA’s Cosmos Curator and AI model fine-tuning microservices.

The initiative addresses a key challenge for AI developers—access to high-quality, regulation-compliant video data for training visual AI models. By leveraging the NVIDIA platform, Project Hafnia will allow data generators to share and transfer datasets, ensuring AI developers can access traceable, high-quality annotated video data for accelerated model training.

According to Milestone Systems, the project will streamline video data movement across industries, significantly improving AI-powered analytics development for traffic monitoring, smart infrastructure, and public safety applications.

AI Model Training as a Service

One of the key components of Project Hafnia is Training as a Service, where software developers can access high-quality, easily transferable video data for AI model training. Additionally, Milestone Systems will introduce a Visual Language Model (VLM) as a Service, fine-tuned for smart city transportation analytics using NVIDIA’s AI fine-tuning microservices and Cosmos Curator.

The VLM is optimised to run on NVIDIA GPUs and will be integrated into NVIDIA’s Video Search and Summarisation (VSS) AI blueprint, improving the efficiency of video-based AI model training and enhancing real-time data accessibility.

“By leveraging the NVIDIA platform, Milestone Systems is helping accelerate this next wave of powerful visual services,” said Deepu Talla, Vice President and General Manager of Embedded and Edge Computing at NVIDIA. He added that the next phase in developing and adopting visually perceptive Agentic AI services will be unlocked by recipes like NVIDIA’s VSS blueprint combined with widely available and accessible fine-tuned VLM models.

Seamless Data Access for AI Developers

Milestone Systems is aggregating compliant video datasets through its global network of partners and customers. These datasets will be used by organisations that need AI-driven analytics for traffic monitoring, manufacturing, airports, law enforcement, and business intelligence.

By improving data annotation, curation, and transfer, the project aims to eliminate bottlenecks in AI training, ensuring faster and more efficient access to video data. The NVIDIA-powered service is expected to accelerate AI model development up to 30 times faster than today’s standards.

AI-Driven Video Data Exchange for Traffic Management

The first application of Project Hafnia is a traffic and transportation-focused Visual Language Model. This model will support traffic assessments, driving condition evaluations, alert validation, and incident reporting, all of which depend on efficient and secure video data transfer.

Thomas Jensen, Chief Executive Officer at Milestone Systems, said streamlined data access is critical for AI development. “Artificial intelligence is our generation’s biggest game-changer. A major challenge for ongoing development is access to high-quality data,” he said adding that the Project Hafnia platform will collect and curate data with the aspiration to be the world’s smartest, fastest, and most responsible platform for video data and AI training.

By leveraging high-quality datasets, regulation-compliant data transfer, and AI model optimisation, Milestone Systems aims to improve video-based AI analytics and enable faster, more accessible AI training for industries relying on video intelligence.