Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Semi Realtime Photo Classification
Search
Sponsored
·
Ship Features Fearlessly
Turn features on and off without deploys. Used by thousands of Ruby developers.
→
Yuichiro Someya
March 30, 2017
Programming
87
0
Share
Semi Realtime Photo Classification
https://jawsug-hpc.connpass.com/event/52263/
Yuichiro Someya
March 30, 2017
More Decks by Yuichiro Someya
See All by Yuichiro Someya
にんげんがさき 基盤はあと / Developers over ML platform
ayemos
0
15k
機械学習をスモールスタートさせる方法 / small machine learning
ayemos
3
2.1k
アットホームな分析基盤の作り方 / Homemade Machine Learning Toolkits
ayemos
1
1k
サービス開発、機械学習、クラウド / the trinity of machine learning
ayemos
0
3.6k
成長を止めない機械学習のやり方 / Don't stop 'til you get enough (data).
ayemos
15
5.3k
AWS で加速する機械学習 / Accelerate Machine Learning with AWS
ayemos
1
360
クックパッドの機械学習基盤 2018 / Machine Learning Platform at Cookpad ~ 2018 ~
ayemos
15
21k
PyTorchとCaffe2とONNXと深層学習モデルのデプロイについて
ayemos
1
3.1k
クックパッドにおけるAWS GPUインスタンスの利用事例 / Powering by AWS GPU Instances in Cookpad Inc
ayemos
0
450
Other Decks in Programming
See All in Programming
TiDBのアーキテクチャから学ぶ分散システム入門 〜MySQL互換のNewSQLは何を解決するのか〜 / tidb-architecture-study
dznbk
1
160
Swift Concurrency Type System
inamiy
0
460
それはエンジニアリングの糧である:AI開発のためにAIのOSSを開発する現場より / It serves as fuel for engineering: insights from the field of developing open-source AI for AI development.
nrslib
1
840
UIの境界線をデザインする | React Tokyo #15 メイントーク
sasagar
1
270
ローカルで稼働するAI エージェントを超えて / beyond-local-ai-agents
gawa
3
270
今こそ押さえておきたい アマゾンウェブサービス(AWS)の データベースの基礎 おもクラ #6版
satoshi256kbyte
1
240
[PHPerKaigi 2026]PHPerKaigi2025の企画CodeGolfが最高すぎて社内で内製して半年運営して得た内製と運営の知見
ikezoemakoto
0
340
ファインチューニングせずメインコンペを解く方法
pokutuna
0
300
アーキテクチャモダナイゼーションとは何か
nwiizo
17
4.9k
条件判定に名前、つけてますか? #phperkaigi #c
77web
2
1k
ルールルルルルRubyの中身の予備知識 ── RubyKaigiの前に予習しなイカ?
ydah
1
140
安いハードウェアでVulkan
fadis
1
950
Featured
See All Featured
Breaking role norms: Why Content Design is so much more than writing copy - Taylor Woolridge
uxyall
0
250
Writing Fast Ruby
sferik
630
63k
A Guide to Academic Writing Using Generative AI - A Workshop
ks91
PRO
1
270
Beyond borders and beyond the search box: How to win the global "messy middle" with AI-driven SEO
davidcarrasco
3
110
Leo the Paperboy
mayatellez
7
1.6k
RailsConf 2023
tenderlove
30
1.4k
Why Mistakes Are the Best Teachers: Turning Failure into a Pathway for Growth
auna
0
120
Docker and Python
trallard
47
3.8k
Digital Ethics as a Driver of Design Innovation
axbom
PRO
1
260
Everyday Curiosity
cassininazir
0
190
The Invisible Side of Design
smashingmag
302
51k
Large-scale JavaScript Application Architecture
addyosmani
515
110k
Transcript
Semi-Realtime Photo Classification Yuichiro Someya @ Cookpad Inc.
Semi-Realtime Photo Classification Yuichiro Someya @ Cookpad Inc.
BZFNPTNF • ࣗݾհ • ʮྉཧ͖Ζ͘ʯͷհ ‣ ΞʔΩςΫνϟ/Architecture ‣ ӡ༻࣮/Results &
Insights Agenda
BZFNPTNF છ୩༔Ұ<:VJDIJSP4PNFZB> UXJUUFS!BZFNPT@Z HJUIVC!BZFNPT TQFBLFSEFDLBZFNPT XFCBZFNPTNF ౦ژۀେֶɹใֶ෦ଔۀ ಉେֶӃɹܭࢉֶઐ߈ɹम࢜ ΫοΫύουגࣜձࣾ ৽ଔݚڀ։ൃ෦
ը૾ೝࣝ ݚڀ։ൃج൫
NFEJVNDPN!BZFNPT
None
ϨγϐສҎ্ ݄ ࣍ ར ༻ ऀ ສ ਓ Ҏ
্
None
None
None
None
None
ྉཧ͖Ζ͘
w εϚʔτϑΥϯͷࣸਅͷɺྉཧΛࣗಈతʹऩूʗه w $//ʹΑΔྉཧը૾ೝࣝ w ਓҎ্ͷϢʔβʔ͕ར༻ w
ຕҎ্ͷྉཧࣸਅ ݱࡏ ྉཧ͖Ζ͘
BZFNPTNF ΞʔΩςΫνϟ
BZFNPTNF w ը૾ॲཧͱఆॲཧʹֻ͕͔ͦͦ࣌ؒ͜͜Δ dNT w "1*αʔόʔ͔Βಉظతʹୟ͚ͳ͍ 6OJDPSOͷXPSLFS͕ਚ͖ͯ͠·͏ w
"NB[PO4 424Λར༻ͨ͠ඇಉظͳఆॲཧϫʔΫϑϩʔ
BZFNPTNF $MJFOU "OESPJE J04 "1*4FSWFS SVCZ $MBTTJpDBUJPO8PSLFS QZUIPO SFTVMU\JT@GPPECPPM^ <6QMPBEQIPUPUPT>
1045JT@QIPUP\LFZ@PO@TTUSJOH^ FORVFVF\LFZ@PO@TTUSJOH^ EFRVFVF \LFZ@PO@TTUSJOH^ 1045SFTVMU \LFZ@PO@TTUSJOH SFTVMU\JT@GPPE <%PXOMPBE*NBHF> %#
BZFNPTNF $MJFOU "OESPJE J04 "1*4FSWFS SVCZ $MBTTJpDBUJPO8PSLFS QZUIPO SFTVMU\JT@GPPECPPM^ <6QMPBEQIPUPUPT>
1045JT@QIPUP\LFZ@PO@TTUSJOH^ FORVFVF\LFZ@PO@TTUSJOH^ EFRVFVF \LFZ@PO@TTUSJOH^ 1045SFTVMU \LFZ@PO@TTUSJOH SFTVMU\JT@GPPE <%PXOMPBE*NBHF> %#
BZFNPTNF $MJFOU "OESPJE J04 "1*4FSWFS SVCZ $MBTTJpDBUJPO8PSLFS QZUIPO SFTVMU\JT@GPPECPPM^ <6QMPBEQIPUPUPT>
1045JT@QIPUP\LFZ@PO@TTUSJOH^ FORVFVF\LFZ@PO@TTUSJOH^ EFRVFVF \LFZ@PO@TTUSJOH^ 1045SFTVMU \LFZ@PO@TTUSJOH SFTVMU\JT@GPPE <%PXOMPBE*NBHF> %# ඇಉظʹఆॲཧ
BZFNPTNF $MJFOU "OESPJE J04 "1*4FSWFS SVCZ $MBTTJpDBUJPO8PSLFS QZUIPO %# "QQSPYJNBUF"HF0G0MEFTU.FTTBHF
4DBMFPVUJO
BZFNPTNF $MBTTJpDBUJPO8PSLFS QZUIPO "QQSPYJNBUF"HF0G0MEFTU.FTTBHF 4DBMFPVUJO
Scale GPU Applications on ECS
BZFNPTNF $MBTTJpDBUJPO8PSLFS QZUIPO "QQSPYJNBUF"HF0G0MEFTU.FTTBHF 4DBMFPVUJO
BZFNPTNF JOTUBODF@ $16 .FNPSZ 5BTL $16 .FNPSZ Α͋͘Δ&$4ͷ4DBMF0VU 5BTL $16
.FNPSZ 5BTL $16 .FNPSZ
BZFNPTNF JOTUBODF@ $16 .FNPSZ 5BTL $16 .FNPSZ Α͋͘Δ&$4ͷ4DBMF0VU 5BTL $16
.FNPSZ 5BTL $16 .FNPSZ JOTUBODF@ $16 .FNPSZ
BZFNPTNF JOTUBODF@ $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN( (16Λ͍ͬͯΔ߹
5BTL $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN(
BZFNPTNF JOTUBODF@ $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN( (16Λ͍ͬͯΔ߹
5BTL $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN(
BZFNPTNF JOTUBODF@ $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN( (16Λ͍ͬͯΔ߹
5BTL $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN(
BZFNPTNF JOTUBODF@ $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN( (16Λ͍ͬͯΔ߹
5BTL $16 .FNPSZ (16.FN( 5BTL $16 .FNPSZ (16.FN( JOTUBODF@ $16 .FNPSZ (16.FN(
ӡ༻ͯ͠Έͯ
BZFNPTNF w େମͳ͍
BZFNPTNF