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
GNeRF: GAN-based Neural Radiance Field without ...
Search
peisuke
December 12, 2021
Technology
1
780
GNeRF: GAN-based Neural Radiance Field without Posed Camera
第9回全日本コンピュータビジョン勉強会
ICCV2021論文読み回発表資料
GNeRF: GAN-based Neural Radiance Field without Posed Camera
peisuke
December 12, 2021
Tweet
Share
More Decks by peisuke
See All by peisuke
AI for Kids:小学生に画像認識を教えてみた話
peisuke
1
29
LangGraphで始めるマルチエージェントシステム
peisuke
13
4.5k
Self-RAG: Learning to Retrieve, Generate and Critique through Self-Reflections
peisuke
9
1.5k
Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fields
peisuke
0
11k
LangChain Toolsの運用と改善
peisuke
5
2.8k
TTS Skins: Speaker Conversion via ASR
peisuke
0
400
A Quantum Computational Approach to Correspondence Problems on Point Sets
peisuke
0
730
F0-Consistent Many-to-many Non-parallel Voice Conversion via Conditional Autoencoder
peisuke
0
210
YOLACT real-time instance segmentation
peisuke
1
300
Other Decks in Technology
See All in Technology
Snowflake Intelligenceで実現できるノーコードAI活用
takumimukaiyama
1
250
OpenTelemetry Collector internals
ymotongpoo
5
550
CSS、JSをHTMLテンプレートにまとめるフロントエンド戦略
d120145
0
100
今からでも間に合う! 生成AI「RAG」再入門 / Re-introduction to RAG in Generative AI
hideakiaoyagi
1
180
SFTPコンテナからファイルをダウンロードする
dip
0
400
「規約、知識、オペレーション」から考える中規模以上の開発組織のCursorルールの 考え方・育て方 / Cursor Rules for Coding Styles, Domain Knowledges and Operations
yuitosato
6
1.9k
What's new in OpenShift 4.19
redhatlivestreaming
1
280
“プロダクトを好きになれるか“も QAエンジニア転職の大事な判断基準だと思ったの
tomodakengo
0
140
堅牢な認証基盤の実現 TypeScriptで代数的データ型を活用する
kakehashi
PRO
2
230
OCI Oracle Database Services新機能アップデート(2025/03-2025/05)
oracle4engineer
PRO
1
160
DroidKnights 2025 - Jetpack XR 살펴보기: XR 개발은 어떻게 이루어지는가?
heesung6701
1
120
上長や社内ステークホルダーに対する解像度を上げて、より良い補完関係を築く方法 / How-to-increase-resolution-and-build-better-complementary-relationships-with-your-bosses-and-internal-stakeholders
madoxten
13
7.8k
Featured
See All Featured
The Invisible Side of Design
smashingmag
299
51k
The Cult of Friendly URLs
andyhume
79
6.4k
Optimizing for Happiness
mojombo
379
70k
[Rails World 2023 - Day 1 Closing Keynote] - The Magic of Rails
eileencodes
35
2.3k
Cheating the UX When There Is Nothing More to Optimize - PixelPioneers
stephaniewalter
281
13k
GraphQLの誤解/rethinking-graphql
sonatard
71
11k
The Web Performance Landscape in 2024 [PerfNow 2024]
tammyeverts
8
660
Sharpening the Axe: The Primacy of Toolmaking
bcantrill
43
2.4k
How to Think Like a Performance Engineer
csswizardry
24
1.7k
It's Worth the Effort
3n
184
28k
Put a Button on it: Removing Barriers to Going Fast.
kastner
60
3.9k
Imperfection Machines: The Place of Print at Facebook
scottboms
267
13k
Transcript
GNeRF: GAN-based Neural Radiance Field without Posed Camera @peisuke 9
ABEJA Twitter @peisuke Github https://github.com/peisuke Qiita https://qiita.com/peisuke SlideShare https://www.slideshare.net/FujimotoKeisuke
• • GNeRF: GAN-based Neural Radiance Field without Posed Camera
• • NeRF • • • GAN Radiance Field • RF
NeRF • / 5 •
• • (x,y,z)T (φ1,θ1)T (φ2,θ2)T (x,y,z) φ2,θ2 (x,y,z) φ1,θ1
NeRF • • •
NeRF • NeRF in the Wild • NeRF • iNeRF
• iNeRF NeRF in the Wild BP
条件 iNeRF NeRF-- IDR GNeRF(提案⼿法) 事前学習 必要 不要 不要 不要
粗いカメラ座標 必要 必要 必要 不要 前⽅のみの制約 不要 必要 不要 不要 マスク 不要 不要 必要 不要 • •
• • • • • GAN Radiance Field • RF
• Phase A Phase B
RF • D E 3
GAN Real • • Discriminator Real • Dynamic Patch Sampling
/ 16x16 • Dynamic Patch Sampling
GAN Fake • • NeRF 16x16 • G
• 6 𝑹 = 𝑟!! 𝑟!" 𝑟!# 𝑟"! 𝑟"" 𝑟"#
𝑟#! 𝑟#" 𝑟## 𝒓 = 𝑟!! 𝑟!" 𝑟!# 𝑟"! 𝑟"" 𝑟"# $ • SO(3) NN • On the Continuity of Rotation Representations in Neural Networks
• Inversion network • Static Patch Sampling • E •
Static Patch Sampling
G G D G
• • Inversion network • refinement
• Refinement
• • • • • Photometric loss
• • Generator • NeRF • Discriminator • Instance Normalization
Spectral Normalization • Inversion network • Vision Transformer • • A • A B 1 iteration • B
• • COLMAP NeRF C+n • Refine C+r • Ground
Truth NeRF G+n • • Synthetic-NeRF image size: 400x400 DTU 500x400 • 100 • • Peak Signal-to-Noise Ratio SSIM LPIPS • / • Single TAITAN GPU 30
COLMAP •
• COLMAP Drums
難しいシーン 特徴の少ない シーン
IDR • 3D IDR • IDR • PNSR •
• COLMAP • Drums Lego COLMAP
Ablation Study • • GAN InvNet. PhaseB • • A
B A ABABABAB B
• • • COLMAP • •
• NeRF • • GAN RF • RF • •
• COLMAP