Skip to main content

Machine Learning training

Machine Learning training

Machine Learning Training

VRSG’s extensive, high-fidelity 3D model library supports AI/Machine Learning training for Large Language Model (LLM) programs.

Under a newly introduced licensing mechanism, VRSG 3D content can now be exported for use in end user applications. For a set fee users can export 3D model images in nearly limitless depictions for direct input into LLM for training sensors for target recognition applications.

VRSG provides limitless datasets of visually accurate 3D military platforms in high-resolution geospecific terrain to train LLMs for intelligence, surveillance and reconnaissance applications such as combat vehicle identification and target recognition. This allows users to create high-volume image datasets of platforms in many different environments to effectively train LLMs to accurately and reliably identify the platform from any given angle.

  • VRSG’s model library includes more than 10,200 3D models, including military platforms such as vehicles, aircraft, maritime vessels, radar, weapon systems and munitions, as well as buildings, characters and cultural entities
  • The library is continually updated with the latest military and civilian platforms deployed in conflicts around the globe in accurate paint schemes
  • Each model entity includes highly-detailed geometry
  • The majority of vehicle, aircraft and munition models include articulated parts, damage states, and advanced animations such as moving wheels or tank tracks that turn at a rate coupled to vehicle velocity
  • Many models support time real-time, physics-based thermal sensor viewing within VRSG
  • VRSG’s configurable environmental effects settings replicate the varied appearance of assets in different light/dark/shadow and weather conditions
  • VRSG’s whole-world terrain database includes high-resolution insets of urban areas and military training sites, enabling recognition training within different real-world settings
  • With scripted camera settings hundreds of platform images can be used to train LLMs.

VRSG users on active software maintenance can access all new and updated models as they are added to the library, at no additional costs. Users can also request new models for inclusion in the library. 

Real vehicle captured at Fallon Training Range by a UAV sensor (left), and a VRSG 3D model entity of the same vehicle type in VRSG's virtual Fallon terrain (right). Notice original image shadow has the sun low and behind the vehicle, and in the VRSG scene the sun is higher and more lateral. A key benefit of using VRSG + 3D Model + Terrain is the ability to set various sun angles to train LLM to better discern vehicle from shadows in aerial sensor footage.

For pricing, please see the price list and request for quote form.