EVA2: Exploiting temporal redundancy in live computer vision

Mark Buckler, Philip Bedoukian, Suren Jayasuriya, Adrian Sampson

Research output: Chapter in Book/Report/Conference proceedingConference contribution

14 Scopus citations

Abstract

Hardware support for deep convolutional neural networks (CNNs) is critical to advanced computer vision in mobile and embedded devices. Current designs, however, accelerate generic CNNs; they do not exploit the unique characteristics of real-time vision. We propose to use the temporal redundancy in natural video to avoid unnecessary computation on most frames. A new algorithm, activation motion compensation, detects changes in the visual input and incrementally updates a previously-computed activation. The technique takes inspiration from video compression and applies well-known motion estimation techniques to adapt to visual changes. We use an adaptive key frame rate to control the trade-off between efficiency and vision quality as the input changes. We implement the technique in hardware as an extension to state-of-the-art CNN accelerator designs. The new unit reduces the average energy per frame by 54%, 62%, and 87% for three CNNs with less than 1% loss in vision accuracy.

Original languageEnglish (US)
Title of host publicationProceedings - 2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture, ISCA 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages533-546
Number of pages14
ISBN (Electronic)9781538659847
DOIs
StatePublished - Jul 19 2018
Event45th ACM/IEEE Annual International Symposium on Computer Architecture, ISCA 2018 - Los Angeles, United States
Duration: Jun 2 2018Jun 6 2018

Publication series

NameProceedings - International Symposium on Computer Architecture
ISSN (Print)1063-6897

Other

Other45th ACM/IEEE Annual International Symposium on Computer Architecture, ISCA 2018
CountryUnited States
CityLos Angeles
Period6/2/186/6/18

Keywords

  • Application specific integrated circuits
  • Computer architecture
  • Computer vision
  • Convolutional neural networks
  • Hardware acceleration
  • Video compression

ASJC Scopus subject areas

  • Hardware and Architecture

Fingerprint Dive into the research topics of 'EVA<sup>2</sup>: Exploiting temporal redundancy in live computer vision'. Together they form a unique fingerprint.

Cite this