Scholarlink

This paper explores the use of Field-Programmable Gate Arrays (FPGAs) to accelerate a Convolutional Neural Network (CNN) for precision agriculture applications using the FINN framework by Xilinx. Our research provides an end-to-end use case from training and quantizing a CNN to running inference on an FPGA. We evaluate the quantized CNN model on a custom dataset for sugar beet and weed classification. The results demonstrate inference-performance improvements, making this FPGA-based implementation a candidate for real-time agricultural monitoring and decision-making.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hardware Acceleration of CNNs with the FINN Framework

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Domenic Drechsel,
  • Stefan Henkler,
  • Sheikh Muhammad Adib Bin Sh Abu Bakar,
  • Kathleen Strodick,
  • Lukas Walter,
  • Kristian Rother,
  • Ali Ehteshami Bejnordi

摘要

This paper explores the use of Field-Programmable Gate Arrays (FPGAs) to accelerate a Convolutional Neural Network (CNN) for precision agriculture applications using the FINN framework by Xilinx. Our research provides an end-to-end use case from training and quantizing a CNN to running inference on an FPGA. We evaluate the quantized CNN model on a custom dataset for sugar beet and weed classification. The results demonstrate inference-performance improvements, making this FPGA-based implementation a candidate for real-time agricultural monitoring and decision-making.