What is it about?
The efficient recognition of handwritten digits deployed on the Zynq embedded platform is achieved through the High-Level Synthesis (HLS) optimiza-tion method. Finally, the relationship between the performance and hardware resource consumption of the two neural networks is com-prehensively compared and analyzed.
Featured Image
Photo by Thomas Jensen on Unsplash
Why is it important?
Neural networks are more and more widely used in industrial produc-tion. Traditional central processing unit (CPU) and graphics processing unit (GPU) neural network deployment platforms have the disad-vantages of large volume and high power consumption. Platforms based on Advanced RISC Machines (ARM) processors, although easy to deploy, suffer from the disadvantage of low computing power.
Perspectives
To solve this problem, based on Zynq embedded platform, this paper makes full use of its Field Programmable Gate Array (FPGA) side par-allel computing characteristics, designs a fully connected neural net-work (FCNN) and a convolutional neural network (CNN).
jin liu
Read the Original
This page is a summary of: Research on Embedded Deployment Optimization Method of Neural Network, April 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3594300.3594313.
You can read the full text:
Contributors
The following have contributed to this page
