FPGA Accelerated Deep Learning for Industrial and Engineering Applications: Optimal Design Under Resource Constraints

被引:0
|
作者
Liu, Yanyi [1 ]
Du, Hang [1 ]
Wu, Yin [1 ]
Mo, Tianli [1 ]
机构
[1] Nanjing Forestry Univ, Coll Informat Sci & Technol, Nanjing 210037, Peoples R China
来源
ELECTRONICS | 2025年 / 14卷 / 04期
关键词
object detection; YOLOv4-Tiny; refined resource management strategy; predefined interface latency; dynamic bit width tuning quantization;
D O I
10.3390/electronics14040703
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
In response to the need for deploying the YOLOv4-Tiny model on resource-constrained Field-Programmable Gate Array (FPGA) platforms for rapid inference, this study proposes a general optimization acceleration strategy and method aimed at achieving fast inference for object detection networks. This approach centers on the synergistic effect of several key strategies: a refined resource management strategy that dynamically adjusts FPGA hardware resource allocation based on the network architecture; a dynamic dual-buffering strategy that maximizes the parallelism of data computation and transmission; an interface access latency pre-configuration strategy that effectively improves data throughput; and quantization operations for dynamic bit width tuning of model parameters and cached variables. Experimental results on the ZYNQ7020 platform demonstrate that this accelerator operates at a frequency of 200 MHz, achieving an average computing performance of 36.97 Giga Operations Per Second (GOPS) with an energy efficiency of 8.82 Giga Operations Per Second per Watt (GOPS/W). Testing with a metal surface defect dataset maintains an accuracy of approximately 90% per image, while reducing the inference delay per frame to 185 ms, representing a 52.2% improvement in inference speed. Compared to other FPGA accelerator designs, the accelerator design strategies and methods proposed in this study showcase significant enhancements in average computing performance, energy efficiency, and inference latency.
引用
收藏
页数:22
相关论文
共 50 条
  • [31] Modeling and analysis of protein design under resource constraints
    Nov, Y
    Wein, LM
    JOURNAL OF COMPUTATIONAL BIOLOGY, 2005, 12 (02) : 247 - 282
  • [32] Managing product design quality under resource constraints
    Reich, Y
    Levy, E
    INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH, 2004, 42 (13) : 2555 - 2572
  • [33] Recent Developments in Deep Learning for Engineering Applications
    Voulodimos, Athanasios
    Doulamis, Nikolaos
    Bebis, George
    Stathaki, Tania
    COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE, 2018, 2018
  • [34] Research on accelerated coding absorber design with deep learning
    Tian, Feng
    Zhang, Shiao
    Cao, Miao
    Huang, Xiaojun
    PHYSICA SCRIPTA, 2023, 98 (09)
  • [35] FA3C: FPGA-Accelerated Deep Reinforcement Learning
    Cho, Hyungmin
    Oh, Pyeongseok
    Park, Jiyoung
    Jung, Wookeun
    Lee, Jaejin
    TWENTY-FOURTH INTERNATIONAL CONFERENCE ON ARCHITECTURAL SUPPORT FOR PROGRAMMING LANGUAGES AND OPERATING SYSTEMS (ASPLOS XXIV), 2019, : 499 - 513
  • [36] Design of optical meta-structures with applications to beam engineering using deep learning
    Singh, Robin
    Agarwal, Anu
    Anthony, Brian W.
    SCIENTIFIC REPORTS, 2020, 10 (01)
  • [37] Design of optical meta-structures with applications to beam engineering using deep learning
    Robin Singh
    Anu Agarwal
    Brian W. Anthony
    Scientific Reports, 10
  • [38] Learning Optimal Decision Trees Under Memory Constraints
    Aglin, Gael
    Nijssen, Siegfried
    Schaus, Pierre
    MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES, ECML PKDD 2022, PT V, 2023, 13717 : 393 - 409
  • [39] Exploration and Tradeoffs of different Kernels in FPGA Deep Learning Applications
    Delaye, Elliott
    Sirasao, Ashish
    Ghasemi, Ehsan
    PROCEEDINGS OF THE 2018 INTERNATIONAL SYMPOSIUM ON PHYSICAL DESIGN (ISPD'18), 2018, : 42 - 47
  • [40] FPGA-based Deep Learning Accelerator for RF Applications
    den Boer, H.
    Muller, R. W. D.
    Wong, S.
    Voogt, V.
    2021 IEEE MILITARY COMMUNICATIONS CONFERENCE (MILCOM 2021), 2021,