Teledyne FLIR, a subsidiary of Teledyne Technologies Incorporated and a leader in advanced sensing solutions, has unveiled its latest technological innovation, the Prism AIMMGen™. This development joins the ranks of the Teledyne FLIR Prism™ software family and offers cutting-edge improvements in automated artificial intelligence (AI) model generation. A key feature of the Prism AIMMGen is its reliance on synthetically generated data, which serves to optimise AI model creation in a more efficient and cost-effective manner.

The Prism AIMMGen™ promises substantial benefits for system integrators involved in developing AI/ML products across various sectors, including commercial enterprises, first response units, and defence applications. The traditional process of manually labelling data for AI and machine learning products is both labour-intensive and costly, especially when the task involves uncommon or complex target objects such as military targets. Collection of this data from real-world environments presents both practical and financial challenges. However, the Prism AIMMGen effectively addresses these issues through its sophisticated synthetic data generation capabilities.

This new tool from Teledyne FLIR can produce millions of machine-learning-ready examples of chosen objects across different conditions and scenarios, utilising both visible and infrared spectrums. By enabling the generation of annotated synthetic imagery quickly, developers can significantly reduce the time and resources typically required for engineering and development processes.

Dan Walker, Vice President of Teledyne FLIR, emphasised the importance of adopting such advanced synthetic data to keep pace with the rapid developments of AI systems. He stated that Prism AIMMGen not only accelerates the time it takes for developers to bring AI products to market but also enhances the preparedness of these technologies to handle diverse scenarios under a wide range of environmental conditions.

The Prism AIMMGen's automated system for AI optimisation is designed to produce high-quality models without the need for expensive real-world data collection. The in-house machine learning operations infrastructure ensures the security of data while leveraging a vast synthetic data lake. This approach allows for cost-effective AI model development and the production of new models quickly, adapting to new task requirements or target classes as needed.

Key features of the Prism AIMMGen include:

  • Access to unlimited synthetic images from a large ITAR-free data lake, designed for various object detection applications.
  • The capacity to produce millions of automatically labelled synthetic image datasets tailored to specific application needs, which ensures focused and efficient AI model training.
  • On-premise automation for model training, validation, and optimisation on secure MLOps cloud infrastructure, offering a timely and efficient performance in AI/ML solutions development.

For further information, interested parties are encouraged to visit Teledyne FLIR’s website or attend the AUSA 2024 Annual Meeting & Exposition from October 14-16 in Washington, D.C.

Teledyne FLIR is recognised globally for its intelligent sensing solutions and operates under the wider umbrella of Teledyne Technologies. The company, established in 1978, employs approximately 4,000 personnel globally and is dedicated to developing advanced technologies that facilitate quicker and more informed decision-making to preserve both lives and assets.

Source: Noah Wire Services