Working with Omnitek for AI
The Omnitek DPU design team has achieved what no other team has been able to achieve by applying its expertise in vision, signal processing, machine learning and optimum FPGA design to achieve the world-leading performance in FPGA based machine learning IP.
The Omnitek DPU is available today, for implementation as an inference engine in a Xilinx UltraScale+ FPGA. After reviewing the specific needs of your application, Omnitek can provide a standard or customised configuration that delivers the optimum balance of performance, cost and power. You are then empowered through the software programming interface to further customise the DPU for your needs and to program it with your training data. In order to protect your investment, even after launching your product, Omnitek can periodically review with you advances in machine learning algorithms which could extend the life of your product or improve its performance, power consumption or cost. Omnitek has developed a portfolio of additional IP for video/vision applications which can be integrated with the DPU to create complete intelligent video/vision systems on a single chip.
If you would like to discuss engaging with Omnitek on your next machine learning design, please contact Omnitek now.
Inference and training implementations on alternative FPGA/SoC platforms are anticipated in the near future. For any of these options, please contact Omnitek for more information.
In order continuously to develop the most efficient and useful machine learning solutions for its customers, Omnitek is working with an ecosystem of specialist partners including the University of Oxford, Cubica Technology and Xilinx.
The University of Oxford is a world-renowned seat of learning and Omnitek is proud to sponsor an Oxford DPhil (PhD) graduate, majoring in research into optimum implementations of machine learning algorithms on FPGAs for inference and training.
The DPhil is co-supervised by Omnitek and Oxford University, specifically the Active Vision Laboratory in the Engineering Science department, which seeks to advance knowledge in computational vision, particularly in the area of 3D scene reconstruction from conventional and depth imagery. In keeping with the spirit of open research, key findings of this DPhil work will be published in academic papers.
Cubica Technology is a company founded, led & teamed by industry-leading experts in machine learning algorithms, with many clients developing security and defence products. Omnitek is working on joint projects using the Omnitek DPU and Cubica’s proprietary machine learning algorithms.
Xilinx is the inventor of the FPGA, hardware programmable SoCs and the ACAP. Omnitek is a Premier Partner in Xilinx’s Alliance Program and this long-established partnership allows access to the latest FPGA and MPSoC technologies, nurturing Omnitek’s deep understanding of these devices.