World’s Biggest Chip Design Technology Users Event Reopens Live in...
Flex Logix® Technologies, Inc., supplier of the most-efficient AI edge inference accelerator and the leading supplier of eFPGA IP, today announced that it has partnered with Roboflow, an end-to-end computer vision solution for developers, to produce edge AI vision models built for the Flex Logix InferX™ platform for a wide variety of applications. This partnership will enable customers to train computer vision models that take advantage of superior AI inference capabilities with high-accuracy, high throughout and low power for complex models such as YOLO object detection models needed for applications such as robotic vision, industrial, security, and retail analytics.
“Customers are seeing a need for alternatives to GPUs in edge inferencing applications,” said Dana McCarty, Vice President of Sales and Marketing for Flex Logix’s Inference Products. “Through our partnership with Roboflow, we can empower these customers to use the Roboflow dataset management and model training technology, which eases model development, and also take advantage of our higher processing efficiency and lower power when compared to common GPU-based solutions.”
“At Roboflow, enabling users to build high quality computer vision models and deploy to a wide range of industry leading hardware options like GPU, and now eFPGA, is essential. We are excited to add support for customers to train models and easily deploy them to the Flex Logix InferX X1 accelerator for advanced edge AI workloads,” said Joseph Nelson, CEO at Roboflow. “The InferX accelerator is able to offer a low power, low cost and high efficiency solution, while still offering a compelling level of performance for inferencing applications.”
The partnership creates a seamless connection between building custom computer vision models and easily deploying them into production. Once models are in production and on a device, customers are able to benefit from an active learning pipeline to continue improving their datasets and model performance for their given outcome.