Increasing the scalability of graph convolution for FPGA-implemented event-based vision

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Wzorek, Piotr, Jeziorek, Kamil, Kryjak, Tomasz, Pinna, Andrea
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909402714341376
author Wzorek, Piotr
Jeziorek, Kamil
Kryjak, Tomasz
Pinna, Andrea
author_facet Wzorek, Piotr
Jeziorek, Kamil
Kryjak, Tomasz
Pinna, Andrea
contents Event cameras are becoming increasingly popular as an alternative to traditional frame-based vision sensors, especially in mobile robotics. Taking full advantage of their high temporal resolution, high dynamic range, low power consumption and sparsity of event data, which only reflects changes in the observed scene, requires both an efficient algorithm and a specialised hardware platform. A recent trend involves using Graph Convolutional Neural Networks (GCNNs) implemented on a heterogeneous SoC FPGA. In this paper we focus on optimising hardware modules for graph convolution to allow flexible selection of the FPGA resource (BlockRAM, DSP and LUT) for their implementation. We propose a ''two-step convolution'' approach that utilises additional BRAM buffers in order to reduce up to 94% of LUT usage for multiplications. This method significantly improves the scalability of GCNNs, enabling the deployment of models with more layers, larger graphs sizes and their application for more dynamic scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2411_04269
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Increasing the scalability of graph convolution for FPGA-implemented event-based vision
Wzorek, Piotr
Jeziorek, Kamil
Kryjak, Tomasz
Pinna, Andrea
Computer Vision and Pattern Recognition
Event cameras are becoming increasingly popular as an alternative to traditional frame-based vision sensors, especially in mobile robotics. Taking full advantage of their high temporal resolution, high dynamic range, low power consumption and sparsity of event data, which only reflects changes in the observed scene, requires both an efficient algorithm and a specialised hardware platform. A recent trend involves using Graph Convolutional Neural Networks (GCNNs) implemented on a heterogeneous SoC FPGA. In this paper we focus on optimising hardware modules for graph convolution to allow flexible selection of the FPGA resource (BlockRAM, DSP and LUT) for their implementation. We propose a ''two-step convolution'' approach that utilises additional BRAM buffers in order to reduce up to 94% of LUT usage for multiplications. This method significantly improves the scalability of GCNNs, enabling the deployment of models with more layers, larger graphs sizes and their application for more dynamic scenarios.
title Increasing the scalability of graph convolution for FPGA-implemented event-based vision
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.04269