Embedded Frame Compression for Energy-Efficient Computer Vision Systems

Li Guo, Dajiang Zhou, Jinjia Zhou, Shinji Kimura

研究成果: Conference contribution

1 引用 (Scopus)

抄録

Computer vision applications are rapidly gaining popularity in embedded systems, which typically involve a difficult trade-off between vision performance and energy consumption under a constraint of real-time processing throughput. Recently, hardware (FPGA and ASIC-based) implementations have emerged that significantly improve the energy efficiency of vision computation. These implementations, however, often involve intensive memory traffic that retains a significant portion of energy consumption at the system level. To address this issue, we present a lossy embedded compression framework to exploit the trade-off between vision performance and memory traffic for input images. Differential pulse-code modulation-based gradient-oriented quantization is developed as the lossy compression algorithm. We also present its hardware design that supports up to 12-scale 1080p@60fps real-time processing. For histogram of oriented gradient-based deformable part models on VOC2007, the proposed framework achieved a 49.6%-60.5% memory traffic reduction at a detection rate degradation of 0.05%-0.34%. For AlexNet on ImageNet, memory traffic reduction achieved up to 60.8% with less than 0.61% classification rate degradation.

元の言語English
ホスト出版物のタイトル2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings
出版者Institute of Electrical and Electronics Engineers Inc.
2018-May
ISBN(電子版)9781538648810
DOI
出版物ステータスPublished - 2018 4 26
イベント2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Florence, Italy
継続期間: 2018 5 272018 5 30

Other

Other2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018
Italy
Florence
期間18/5/2718/5/30

Fingerprint

Computer vision
Data storage equipment
Energy utilization
Differential pulse code modulation
Hardware
Degradation
Application specific integrated circuits
Processing
Embedded systems
Energy efficiency
Field programmable gate arrays (FPGA)
Throughput

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

これを引用

Guo, L., Zhou, D., Zhou, J., & Kimura, S. (2018). Embedded Frame Compression for Energy-Efficient Computer Vision Systems. : 2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings (巻 2018-May). [8351483] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISCAS.2018.8351483

Embedded Frame Compression for Energy-Efficient Computer Vision Systems. / Guo, Li; Zhou, Dajiang; Zhou, Jinjia; Kimura, Shinji.

2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings. 巻 2018-May Institute of Electrical and Electronics Engineers Inc., 2018. 8351483.

研究成果: Conference contribution

Guo, L, Zhou, D, Zhou, J & Kimura, S 2018, Embedded Frame Compression for Energy-Efficient Computer Vision Systems. : 2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings. 巻. 2018-May, 8351483, Institute of Electrical and Electronics Engineers Inc., 2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018, Florence, Italy, 18/5/27. https://doi.org/10.1109/ISCAS.2018.8351483
Guo L, Zhou D, Zhou J, Kimura S. Embedded Frame Compression for Energy-Efficient Computer Vision Systems. : 2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings. 巻 2018-May. Institute of Electrical and Electronics Engineers Inc. 2018. 8351483 https://doi.org/10.1109/ISCAS.2018.8351483
Guo, Li ; Zhou, Dajiang ; Zhou, Jinjia ; Kimura, Shinji. / Embedded Frame Compression for Energy-Efficient Computer Vision Systems. 2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings. 巻 2018-May Institute of Electrical and Electronics Engineers Inc., 2018.
@inproceedings{f2ba19ea2f75473ca42e36207d4111fc,
title = "Embedded Frame Compression for Energy-Efficient Computer Vision Systems",
abstract = "Computer vision applications are rapidly gaining popularity in embedded systems, which typically involve a difficult trade-off between vision performance and energy consumption under a constraint of real-time processing throughput. Recently, hardware (FPGA and ASIC-based) implementations have emerged that significantly improve the energy efficiency of vision computation. These implementations, however, often involve intensive memory traffic that retains a significant portion of energy consumption at the system level. To address this issue, we present a lossy embedded compression framework to exploit the trade-off between vision performance and memory traffic for input images. Differential pulse-code modulation-based gradient-oriented quantization is developed as the lossy compression algorithm. We also present its hardware design that supports up to 12-scale 1080p@60fps real-time processing. For histogram of oriented gradient-based deformable part models on VOC2007, the proposed framework achieved a 49.6{\%}-60.5{\%} memory traffic reduction at a detection rate degradation of 0.05{\%}-0.34{\%}. For AlexNet on ImageNet, memory traffic reduction achieved up to 60.8{\%} with less than 0.61{\%} classification rate degradation.",
keywords = "computer vision, embedded compression, energy-efficient, vision-oriented",
author = "Li Guo and Dajiang Zhou and Jinjia Zhou and Shinji Kimura",
year = "2018",
month = "4",
day = "26",
doi = "10.1109/ISCAS.2018.8351483",
language = "English",
volume = "2018-May",
booktitle = "2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Embedded Frame Compression for Energy-Efficient Computer Vision Systems

AU - Guo, Li

AU - Zhou, Dajiang

AU - Zhou, Jinjia

AU - Kimura, Shinji

PY - 2018/4/26

Y1 - 2018/4/26

N2 - Computer vision applications are rapidly gaining popularity in embedded systems, which typically involve a difficult trade-off between vision performance and energy consumption under a constraint of real-time processing throughput. Recently, hardware (FPGA and ASIC-based) implementations have emerged that significantly improve the energy efficiency of vision computation. These implementations, however, often involve intensive memory traffic that retains a significant portion of energy consumption at the system level. To address this issue, we present a lossy embedded compression framework to exploit the trade-off between vision performance and memory traffic for input images. Differential pulse-code modulation-based gradient-oriented quantization is developed as the lossy compression algorithm. We also present its hardware design that supports up to 12-scale 1080p@60fps real-time processing. For histogram of oriented gradient-based deformable part models on VOC2007, the proposed framework achieved a 49.6%-60.5% memory traffic reduction at a detection rate degradation of 0.05%-0.34%. For AlexNet on ImageNet, memory traffic reduction achieved up to 60.8% with less than 0.61% classification rate degradation.

AB - Computer vision applications are rapidly gaining popularity in embedded systems, which typically involve a difficult trade-off between vision performance and energy consumption under a constraint of real-time processing throughput. Recently, hardware (FPGA and ASIC-based) implementations have emerged that significantly improve the energy efficiency of vision computation. These implementations, however, often involve intensive memory traffic that retains a significant portion of energy consumption at the system level. To address this issue, we present a lossy embedded compression framework to exploit the trade-off between vision performance and memory traffic for input images. Differential pulse-code modulation-based gradient-oriented quantization is developed as the lossy compression algorithm. We also present its hardware design that supports up to 12-scale 1080p@60fps real-time processing. For histogram of oriented gradient-based deformable part models on VOC2007, the proposed framework achieved a 49.6%-60.5% memory traffic reduction at a detection rate degradation of 0.05%-0.34%. For AlexNet on ImageNet, memory traffic reduction achieved up to 60.8% with less than 0.61% classification rate degradation.

KW - computer vision

KW - embedded compression

KW - energy-efficient

KW - vision-oriented

UR - http://www.scopus.com/inward/record.url?scp=85051277735&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85051277735&partnerID=8YFLogxK

U2 - 10.1109/ISCAS.2018.8351483

DO - 10.1109/ISCAS.2018.8351483

M3 - Conference contribution

AN - SCOPUS:85051277735

VL - 2018-May

BT - 2018 IEEE International Symposium on Circuits and Systems, ISCAS 2018 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -