ISCAS 2022
|
Technical Lecture and Demo
ID 1452

A Heterogeneous FPGA-Based Accelerator Design for Efficient and Low-Cost Point Clouds Deep Learning Inference
Details
Presenter(s)
Jinling Xu Headshot
Jinling Xu
Affiliation
Beijing Institute of Technology
Country
View profile
Author(s)
Jinling Xu
Affiliation
Beijing Institute of Technology
View profile
Yonggui Wang
Affiliation
Zhejiang Qianjiang Robot Co
View profile
Wenbiao Zhou
Affiliation
Beijing Institute of Technology
View profile
Abstract

In this work, we propose a heterogeneous solution for PointNet++ inference on FPGA. Our work\'s performance density on FPGA outperforms previous works. The flexible on design scale and adaptation to arbitrary network depth make the design suits for different devices and tasks.

Slides
  • A Heterogeneous FPGA-Based Accelerator Design for Efficient and Low-Cost Point Clouds Deep Learning Inference (application/pdf)
    Download