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
三次元形状とディープラーニング
Search
Tatsuya Yatagawa
December 21, 2020
Technology
0
210
三次元形状とディープラーニング
三次元形状処理とディープラーニングの初歩についてまとめたスライドです。
Tatsuya Yatagawa
December 21, 2020
Tweet
Share
More Decks by Tatsuya Yatagawa
See All by Tatsuya Yatagawa
Event-based Camera Simulation using Monte Carlo Path Tracing with Adaptive Denoising (ICIP 2023)
tatsy
0
370
VC'23 チュートリアル講演 『三次元メッシュと深層学習』
tatsy
0
140
SIGGRAPH 2022: Neural Geometry Processing
tatsy
0
160
Real-time Shading with Free-form Planar Area Lights using Linearly Transformed Cosines (JCGT & I3D 2022)
tatsy
0
800
SIGGRAPH Asia 2021: Real-Time Rendering
tatsy
0
54
SIGGRAPH 2021: Denoising and Guiding
tatsy
0
43
SIGGRAPH 2021: Shape Generation
tatsy
0
94
SIGGRAPH 2021: Meshing
tatsy
0
65
12. Keyboard & Mouse
tatsy
0
160
Other Decks in Technology
See All in Technology
終了の危機にあった15年続くWebサービスを全力で存続させる - phpcon2024
yositosi
0
280
Snowflake女子会#3 Snowpipeの良さを5分で語るよ
lana2548
0
230
どちらを使う?GitHub or Azure DevOps Ver. 24H2
kkamegawa
0
690
第3回Snowflake女子会_LT登壇資料(合成データ)_Taro_CCCMK
tarotaro0129
0
180
開発生産性向上! 育成を「改善」と捉えるエンジニア育成戦略
shoota
1
270
Amazon SageMaker Unified Studio(Preview)、Lakehouse と Amazon S3 Tables
ishikawa_satoru
0
150
社内イベント管理システムを1週間でAKSからACAに移行した話し
shingo_kawahara
0
180
ハイテク休憩
sat
PRO
2
140
20241214_WACATE2024冬_テスト設計技法をチョット俯瞰してみよう
kzsuzuki
3
440
組織に自動テストを書く文化を根付かせる戦略(2024冬版) / Building Automated Test Culture 2024 Winter Edition
twada
PRO
13
3.6k
小学3年生夏休みの自由研究「夏休みに Copilot で遊んでみた」
taichinakamura
0
150
MLOps の現場から
asei
6
630
Featured
See All Featured
Easily Structure & Communicate Ideas using Wireframe
afnizarnur
191
16k
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
PRO
10
810
Optimizing for Happiness
mojombo
376
70k
Raft: Consensus for Rubyists
vanstee
137
6.7k
Why Our Code Smells
bkeepers
PRO
335
57k
Rebuilding a faster, lazier Slack
samanthasiow
79
8.7k
Responsive Adventures: Dirty Tricks From The Dark Corners of Front-End
smashingmag
251
21k
Rails Girls Zürich Keynote
gr2m
94
13k
Bash Introduction
62gerente
608
210k
Writing Fast Ruby
sferik
628
61k
A Tale of Four Properties
chriscoyier
157
23k
Imperfection Machines: The Place of Print at Facebook
scottboms
266
13k
Transcript
None
• 機械学習 ニューラルネット 深層学習
• • • θ • = ; = 2 2
+ 1 + 0 ෩ = arg min =0 −1 ( , ; )
• • • 0 = 0 + 0 1 =
1 + 1 2 = 2 + 2 = 0 0 + 1 1 + 2 2 + = 0 0 + 1 1 + 2 2 +(0 0 + 1 1 + 2 2 + )
• • • 0 1 2 0 = ReLU(0 +
0 ) 1 = ReLU(1 + 1 ) 2 = ReLU(2 + 2 )
= 32 = 128 = 512 • •
• • = 512 = 1 = = 4
•
•
•
•
•
•
• • • • • • • • • •
•
(c) Velodyne (c) Andrew Tallon (c) Microsoft (c) Sony
• , , = 0
′ = − ′ = + ′
• ☺ ☺ ☺ ☺ ,
, = 0
☺ ☺ ☺ ☺ , ,
= 0
• • • • •
• • • •
•
•
• ⚫ ⚫ ⚫
• • •
• • •
• •
• • •
• • • • • • •
☺ ☺ ☺ ☺ , ,
= 0
• • • (c) Velodyne
• • • (0 , 0 , 0 ) (1
, 1 , 1 ) ( , , ) (−1 , −1 , −1 ) (+1 , +1 , +1 ) (0 , 0 , 0 ) (1 , 1 , 1 ) (−1 , −1 , −1 ) (+1 , +1 , +1 ) ( , , )
• • •
None
⚫
⚫
• •
• • •
•
•
•
• • • • • • •
☺ ☺ ☺ ☺ , ,
= 0
• • •
• • • •
• • ∗ = ℱ−1(ℱ ⋅ ℱ ) ′ =
ℱ−1(Θ ∘ ℱ )
• = 0 1 1 0 1 0 1 0
1 1 0 1 0 0 1 0 = − − 1 2− 1 2 = :
• • • = T
• () ′ = ℱ−1(Θ ∘ ℱ ) ′ =
ℱ = () ℱ−1 = ()
• • • +1 +2 +1 = ReLU ⋅
• • •
• • •
• •
• • •
• • •
• •
• • • • • •
☺ ☺ ☺ ☺ , ,
= 0
• , , =
• • •
• • • •
• •
• • •
• • •
• • • •
•
• • • •
• • •
• • • •
• • •
• • • •
• • • • • • • •
• • • • • • • • •