Â鶹ÉçÇø

  Abstract: Artificial neural network (ANN) empowered machine vision in embedded platforms has huge potential to be used in clinic environments for medical diagnostics. However, cost-effective artificial neural networks implementation, especially for inference-based classification tasks on assistive diagnostic applications, have suffered unreliable classification performance due to the reduced size of network and precision on embedded platforms.
This project aims to develop an open stack development platform SYCL for supporting open hardware RISC-V in ANN-based machine vision. The benefits of this project are: evaluating state of art machine vision algorithms powered with artificial neural networks for open hardware platforms; providing integration of optimisation of ANN machine vision to Codeplay tool chain; opening a new market of tool chain support for medical devices.