Milestone Systems has announced the launch of Project Hafnia, a platform designed to enhance AI model training by providing access to high-quality, compliant video data. The initiative leverages NVIDIA Cosmos Curator and AI model fine-tuning microservices to facilitate data sharing and utilisation among developers and data generators worldwide.
Project Hafnia aims to be a leading service in AI model training by integrating the latest advances in Vision Language Models (VLM) and data curation capabilities. One of its initial offerings is a VLM fine-tuned using NVIDIA fine-tuning microservices on a vast dataset of compliant transportation data, curated through NVIDIA Cosmos Curator. The platform optimizes accuracy and performance for use on NVIDIA GPUs and within NVIDIA’s Video Search and Summarization (VSS) AI blueprint.
According to Deepu Talla, Vice President and General Manager of Embedded and Edge Computing at NVIDIA, Milestone Systems is playing a crucial role in accelerating the next wave of visual services. He noted that the combination of NVIDIA’s VSS blueprint and fine-tuned VLM models would drive the development and adoption of visually perceptive Agentic AI services.
Milestone Systems, recognized as a global leader in data-driven video technology and analytics, is aggregating compliant data through its extensive network of partners and customers. These organizations are looking to harness their data for smarter analytics, and the new platform will enable them to train AI models with traceable and regulatory-compliant annotated video data.
Thomas Jensen, CEO of Milestone Systems, emphasized that artificial intelligence is a transformative technology, with access to high-quality training data being a key challenge. He described Project Hafnia as an initiative to create the world’s smartest, fastest, and most responsible platform for video data and AI model training.
The platform aims to streamline data access and reduce development time for AI and analytics by up to 30 times compared to current standards. Driven by NVIDIA Cosmos Curator, the service will offer two primary components:
- Training as a Service: A high-quality video data training service that allows software developers to train AI models efficiently.
- Visual Language Model as a Service: A specialized solution for smart city transportation and Intelligent Traffic Services, setting new performance benchmarks in the industry.
With advancements in high-quality data and annotation, Project Hafnia is expected to enhance analytics for various applications, including traffic management, manufacturing, airports, law enforcement, and business operations. The improved accuracy could enable large-scale automation of surveillance and operational processes.
As part of its initial rollout, Milestone Systems is focusing on developing a market-leading traffic and transportation Visual Language Model. This model is designed to support multiple use cases, such as general traffic assessments, driving condition evaluations, alert validation, and incident reporting.
Project Hafnia is launching as a pilot, with developers invited to sign up for early access at https://hafnia.milestonesys.com/joinwaitlist. The initial phase will concentrate on traffic-related video data, with plans to expand its scope once fully operational. Milestone Systems will retain full ownership and management of the platform.