Acceleration of server-side image processing by client-side pre-processing in web application environment

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F62156489%3A43110%2F19%3A43916303" target="_blank" >RIV/62156489:43110/19:43916303 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/00216305:26210/19:PU133026

Výsledek na webu

<a href="http://dx.doi.org/10.1109/TSP.2019.8768889" target="_blank" >http://dx.doi.org/10.1109/TSP.2019.8768889</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/TSP.2019.8768889" target="_blank" >10.1109/TSP.2019.8768889</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Acceleration of server-side image processing by client-side pre-processing in web application environment

Popis výsledku v původním jazyce

The goal of this paper is to verify whether the use of client-side TensorFlow.js, a WebGL-accelerated JavaScript library for machine learning, can accelerate the processing of common photos by computer vision cloud services, such as detection and recognition of specific features like age, sex, expression or specific people in the image. This acceleration is based on pre-processing the input image, namely detecting human faces, which greatly changes the amount of input data that need to be uploaded to the cloud service and thus the amount of uploaded data compared to the original photograph. The upload speed of Internet connection often is, in the case of computer vision cloud services, the bottleneck of the whole system. That&apos;s why decreasing the amount of uploaded data in time shorter than the difference between the total of upload and cloud service processing time of the original and the pre-processed image leads to acceleration.

Název v anglickém jazyce

Acceleration of server-side image processing by client-side pre-processing in web application environment

Popis výsledku anglicky

The goal of this paper is to verify whether the use of client-side TensorFlow.js, a WebGL-accelerated JavaScript library for machine learning, can accelerate the processing of common photos by computer vision cloud services, such as detection and recognition of specific features like age, sex, expression or specific people in the image. This acceleration is based on pre-processing the input image, namely detecting human faces, which greatly changes the amount of input data that need to be uploaded to the cloud service and thus the amount of uploaded data compared to the original photograph. The upload speed of Internet connection often is, in the case of computer vision cloud services, the bottleneck of the whole system. That&apos;s why decreasing the amount of uploaded data in time shorter than the difference between the total of upload and cloud service processing time of the original and the pre-processed image leads to acceleration.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

Výsledek vznikl pri realizaci vícero projektů. Více informací v záložce Projekty.

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2019

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

42nd International Conference on Telecommunications and Signal Processing (TSP)

ISBN

978-1-72811-864-2

ISSN

—

e-ISSN

—

Počet stran výsledku

4

Strana od-do

127-130

Název nakladatele

IEEE-Institute of Electrical Electronics Engineers

Místo vydání

Piscataway

Místo konání akce

Budapešť

Datum konání akce

1. 7. 2019

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000493442800028