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
Summary: Objects as Points
Search
fam_taro
April 22, 2019
Technology
3.3k
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
Summary: Objects as Points
fam_taro
April 22, 2019
More Decks by fam_taro
See All by fam_taro
NeRFの概要と 派生系についてのふんわり紹介
fam_taro
3
4.4k
実践 PyTorch Lightning (2019/11/30 分析コンペLT会 #1)
fam_taro
3
4.6k
Paper:ShapeMask
fam_taro
0
92
Paper-Survey: Objects as Points
fam_taro
0
2.4k
Tensorコアを使った PyTorch の高速化について
fam_taro
4
4.1k
Sequence to Sequence Learning with Neural Networks
fam_taro
1
1.1k
Other Decks in Technology
See All in Technology
AI時代の強いチームの作り方
yuukiyo
26
16k
MIRU 2026 チュートリアル
keisuke198619
0
870
強化学習「理論」入門
enakai00
3
3.6k
【CEDEC2026】コードレビュー支援ツール開発から学ぶ:LLMを用いた業務システムの実践的な運用設計と誤出力対策
cygames
PRO
0
630
生成 AI の基礎 〜 サンプル実装で学ぶ基本原理
enakai00
7
4.4k
【CEDEC2026】『GRANBLUE FANTASY: Relink - Endless Ragnarok』のバトル制作事例 ~最高のキャラゲーを目指して~
cygames
PRO
0
230
名刺メーカーDevグループ 紹介資料
sansan33
PRO
0
1.2k
ブラウザ研修 2026
recruitengineers
PRO
5
800
サイバー捜査員研修(後半)
nomizone
1
810
サイバー捜査員研修(前半)
nomizone
1
1.9k
今こそ聞きたいソフトウェア設計 ドメイン駆動設計再入門
masuda220
PRO
17
6.8k
PLATEAU で バーチャル花火大会
tatsuya1970
0
110
Featured
See All Featured
Design of three-dimensional binary manipulators for pick-and-place task avoiding obstacles (IECON2024)
konakalab
0
530
The AI Revolution Will Not Be Monopolized: How open-source beats economies of scale, even for LLMs
inesmontani
PRO
3
3.7k
Deep Space Network (abreviated)
tonyrice
0
250
Lessons Learnt from Crawling 1000+ Websites
charlesmeaden
PRO
1
1.5k
Statistics for Hackers
jakevdp
799
230k
Ruling the World: When Life Gets Gamed
codingconduct
0
290
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
12
1.8k
How to Get Subject Matter Experts Bought In and Actively Contributing to SEO & PR Initiatives.
livdayseo
0
170
Optimising Largest Contentful Paint
csswizardry
37
3.9k
Why Mistakes Are the Best Teachers: Turning Failure into a Pathway for Growth
auna
0
200
Designing for Performance
lara
611
70k
The Director’s Chair: Orchestrating AI for Truly Effective Learning
tmiket
1
260
Transcript
%BUB%SJWFO%FWFMPQFS.FFUVQ ʲ൪֎ฤ ͖ͳจʹ͍ͭͯޠΔձʳ 0CKFDUTBT1PJOUT $FOUFS/FU IUUQTBSYJWPSHBCT !GBN@UBSP
"HFOEB ͳΜͰ͜ͷจબΜͩͷ ͜Ε·Ͱͷݕग़Ϟσϧͷ՝ $FOUFS/FU ͷಛ ͜ͷจͷ
ݸਓతͳײ
ͳΜͰ͜ͷจબΜͩͷ ࠷ۙग़͔ͨΒʂ ݄ ͪͳΈʹཌʹ l$FOUFS/FU0CKFDU%FUFDUJPOXJUI,FZQPJOU 5SJQMFUTz ͱ͍͏ಉ໊ͷωοτϫʔΫͷจ͕ग़͍ͯΔ ͘͢͝·͗ΘΒΘ͍͠
IUUQTBSYJWPSHBCT ͪ͜Β 405"ͳͷͰͦͷ͏ͪจಡΜͰ͓͖͍ͨ ݕग़ϞσϧͷதͰͯ͘ਫ਼ߴ͍ʂ ϦΞϧλΠϜͳݕग़Λߦ͏্Ͱεϐʔυେࣄ '14 NT ݸਓతʹཉ͍͠ͱ͜Ζ /.4 OPONBYJNVNTVQQSFTTJPO ͕ෆཁʂ ໘Ͱخ͍͠ ޙॲཧ͕ݮͬͨͷ୯७ʹخ͍͠ ͯ͘ɺਫ਼ߴ͍ʂʂ େࣄͳ͜ͱͳͷ̳ʢ :0-0WΑΓ ѹతʹڧ͍ʂ
͜Ε·Ͱͷݕग़Ϟσϧͷ՝ 0CKFDUEFUFDUJPOCZSFHJPODMBTTJGJDBUJPO 3$//ͱ͔ 'BTU3$// େମ TUBHFEFUFDUPS
՝͍ 0CKFDUEFUFDUJPOXJUIJNQMJDJUBODIPST ҉తʹΞϯΧʔΛ͏ͷ େମ TUBHFEFUFDUPS࠷ۙͷ͍ݕग़Ϟσϧ͜Ε 44% 3FUJOB/FU :0-0 W .%FUʜ ީิϘοΫεΛ༧Ί༻ҙ͢Δͷ ϘοΫεղ૾ɾॎԣൺ ՝(SPVOEUSVUI ڭࢣϥϕϧ ࡞࣌ʹɺPWFSMBQͷॏͳΓ۩߹Λ͏ ਓ͕ܾΊ͖͍ͨ͠Ͱ GPSFHSPVOEPSOPUΛܾΊΔ ମʹର͠ෳͷ (SPVOEUSVUI͕͋Γ͏Δ BODIPSʹ͍ͭͯਓ͕ௐ͢Δඞཁ͕͋Δ ݸਓత՝ /.4͕ඞཁ େମNTඞཁ ग़ྗϘοΫε͕ଟ͍ͱ͕͔͔࣌ؒΔ ͕࣌ؒݻఆ͡Όͳ͍
͜Ε·Ͱͷݕग़Ϟσϧͷ՝ 0CKFDUEFUFDUJPOXJUIJNQMJDJUBODIPST ҉తʹΞϯΧʔΛ͏ͷ ՝(SPVOEUSVUI ڭࢣϥϕϧ ࡞࣌ʹɺPWFSMBQͷॏͳΓ۩߹Λ͏
͜Ε·Ͱͷݕग़Ϟσϧͷ՝ 0CKFDUEFUFDUJPOCZLFZQPJOU FTUJNBUJPO IFBUNBQ֤ͬͯΛٻΊΔํ๏ ޙड़ͷ $FOUFS/FU
ͷਪ࣌ͷྲྀΕΛݟΔͱΠϝʔδ͔ͭΈ͍͔͢ $PSOFS/FUϘοΫεͷࠨ্ͱӈԼͷͷΈਪఆ &YUSFNF/FUத৺ͱϘοΫεͷ֯Λਪఆ ՝LFZQPJOU EFUFDUJPOޙʹΈ߹ΘͤΛ HSPVQJOH͢Δඞཁ͕͋Δ ͍
$FOUFS/FU ͷಛ #BDLCPOFͱͯ͠ %-" )PVSHMBTTΛར༻ $PSOFS/FU Ͱར༻͞Ε͍ͯΔ
%-"%FFQMBZFSTBHHSFHBUJPO IUUQTBSYJWPSHQEGQEG %FUFDUJPOʹ͓͍ͯҎԼͭͷΈٻΊΔ ମͷத৺Ґஔ ϘοΫεαΠζ ਤࠨ͕ݩ%-"ɺӈ͕$FOUFS/FU༻%-"
$FOUFS/FU ͷಛ %-"ิ %-"ͷఏҊ ͦͷ·· MPXMBZFSͷΛ͢ͷͰͳ͘ɺͯ͠ΈΑ͏
ԼهݩจΑΓҾ༻
$FOUFS/FU ͷಛ ਪ ༧ଌ ࣌ͷྲྀΕ ը૾Λ #BDLCPOFωοτϫʔΫ %-"
)PVSHMBTT ʹೖྗ ,FZQPJOUIFBUNBQΛग़ྗ ԼهΠϝʔδਤ จதͰ ݩը૾αΠζ )Y8 ʹରͯ͠ ) 8 ͷαΠζ YͳΒ IFBUNBQͷαΠζ Y ͕ࡉ͔͍ MPDBMQFBLΛநग़ ͋ΔՕॴʹ͍ͭͯɺपลݸͷΑΓେ͖͍ Ҏ্ ͱͳΔॴ ͦ͜Λମͷத৺ͱ͢Δʂ MPDBMQFBLͰͷϘοΫεαΠζͱࢄԽޡࠩΛ༧ଌ ࢄԽޡࠩ ݩը૾͔Β IFBUNBQʹͨ͠ࡍͷޡࠩ ඍௐ ༧ଌϘοΫεग़ྗʂ /.4͠ͳ͍ ݸਓతʹେࣄ IFBUNBQͷϚεʹର͢Δग़ྗݸ ݕग़࣌ Ϋϥε ϘοΫεαΠζɺࢄԽޡࠩ IFBUNBQΠϝʔδ https://pythonspot.com/tag/heatmap/
$FOUFS/FU ͷಛ ਪ ༧ଌ ࣌ͷΠϝʔδ λεΫ͝ͱͷҧ͍
$FOUFS/FU ͷಛ 5FTU5JNF"VHNFOUBUJPOͰݕূࡁ /P"VHNFOUBUJPO GMJQ"VHNFOUBUJPO GMJQBOENVMUJTDBMF
XJUI/.4 ˡେࣄ ϦΞϧλΠϜͱͯ͠͏ͳΒ͍Օॴ͕ਫ਼ɾ໘Ͱྑͦ͞͏ #BDLCPOF%-" "VHNFOUBUJPO/PPSGMJQ NVMUJTDBMFਫ਼্͕Δ͚Ͳਪ͕͖͍࣌ؒͭ ίϯϖͳΒ͏Ձ͋Γ͔
$FOUFS/FU ͷಛ ֶश࣌ͷྲྀΕ ڭࢣϥϕϧ༻ IFBUNBQͷ࡞Γํ ॲཧલ
ॲཧޙ ࣮ࡍͷମͷத৺ Λղ૾Խ ຊจͰ 3 ΨγΞϯΧʔωϧͰͳͩΒ͔ͳʹ͢Δ த৺࠲ඪ͕ ͱͳΔࢁΛ࡞ΔΠϝʔδ ಉ͡Ϋϥεͷҧ͏ମͰॏͳͬͨ߹͕ߴ͍ํΛ࠾༻ МQ PCKFDUTJ[FBEBQUJWFTUBOEBSEEFWJBUJPO ଟσʔληοτຖʹग़͖͢ʁ
$FOUFS/FU ͷಛ ֶश࣌ͷ MPTTઃܭ ͦΕͧΕൺΛ͔͚ͯͨ͠ͷ͕߹ܭ MPTT )FBUNBQͷ MPTT
QJYFMXJTFMPHJTUJDSFHSFTTJPOXJUIGPDBMMPTT / OVNCFSPGLFZQPJOUT JOJNBHF ࢄԽޡࠩͷ MPTT ϘοΫεαΠζͷ MPTT ϘοΫεαΠζ TDBMJOH ࠷େʹ͢Δ ͯ͠ͳ͍͜ͱʹҙ
͜ͷจͷ #BDLCPOF %-"ɾ)PVS(MBTT Λ͏ ࢄԽޡࠩͷఆٛ ମͷத৺ͷΈٻΊΕྑ͍ͱ͍͏γϯϓϧͳωοτϫʔΫ
ଞλεΫͷԠ༻ ͕؆୯ͳωοτϫʔΫΛఏҊ %EFUFDUJPO )VNBOQPTFFTUJNBUJPO 0SJFOUBUJPO ํ #BUDIOPSNBMJ[BUJPOͰ DVEOO ΛΘͳ͍ 1Z5PSDIͷιʔεΛ͍ͬͯ͡ઃఆมߋͰ͖Δ ࣮ݧ࣌Θͳ͍ํ͕ਫ਼্͕͕ͬͨͱͷ͜ͱ
ݸਓతͳॴײ #BDLCPOF %-"ɾ)PVS(MBTT ͷߩݙ͕͍͢͝ ࠓޙͬͱ૿͑Δͱࢥ͍·͢ )FBUNBQ࿏ઢͷ %FUFDUPSͬͱ૿͑ͦ͏ʁ
/.4ΛΘͳͯ͘ྑ͍ͷݸਓతʹخ͍͠ γϯϓϧ γϯϓϧͰਫ਼ߴ͍ͷͰ;͍ͭ͘͠ʜ ࣮͕ͱͯࢀߟʹͳΔ ͱΓ͍͋͑ͣۙ͏ͪʹݕূ͠·͢
3FGFSFODFT 0CKFDUTBT1PJOUT IUUQTBSYJWPSHBCT $PSOFS/FU%FUFDUJOH0CKFDUTBT1BJSFE,FZQPJOUT IUUQTBSYJWPSHBCT %FFQ-BZFS"HHSFHBUJPO IUUQTBSYJWPSHBCT
%FFQMBZFSBHHSFHBUJPO$WQS Θ͔Γ͔ͬͨ͢·ͱΊεϥΠυ IUUQTXXXTMJEFTIBSFOFU4IJOJDIJSP.VSBLBNJEFFQMBZFSBHHSFHBUJPODWQS
͓ΘΓ ͋Γ͕ͱ͏͍͟͝·ͨ͠