SlideShare ist ein Scribd-Unternehmen logo
1 von 14
Downloaden Sie, um offline zu lesen
Copyright © 2018 by AI4quant [TW] All rights reserved
AI4quant: AI智慧影像生成
PIXNET Hackathon 2018
Jason Chuang、呂沂樺、Kiki
2018-08-19
Copyright © 2018 by AI4quant [TW] All rights reserved
What is AI?
Big Data
Algorithm
Computing Power
Copyright © 2018 by AI4quant [TW] All rights reserved
Data preprocessing for training data
We use 5 different datasets for training
If the shortest width of the image is less than 256 pixels, ignore
Shrink the shortest width to 256 pixels, cut the center part of the image
Copyright © 2018 by AI4quant [TW] All rights reserved
Big data
Food-11: https://mmspg.epfl.ch/food-image-datasets
DataSet Size(jpg,
png)
Number of
Files
# Training
files
Training
size(bmp,
256x256)
# Validation
files
dataset100 1GB 14,467 10,044 1.83GB 0
dataset256 4GB 31,653 27,002 5GB 0
Food-11 1.1GB 16,643 16,635 3.04GB 0
food-101 4.76GB 101,008 100,956 18.4GB 0
pixfood20 2.03GB 31,305 19,322 3.53GB 19,322
Copyright © 2018 by AI4quant [TW] All rights reserved
Algorithms are motivated by research paper
DCNN based method
Image Inpainting for Irregular Holes Using Partial Convolutions, 2018, Guilin Liu,
Fitsum A. Reda, Kevin J. Shih, Ting-Chun Wang, Andrew Tao, Bryan Catanzaro
High-Resolution Image Inpainting using Multi-Scale Neural Patch Synthesis, 2017,
Chao Yang, Xin Lu,Zhe Lin, Eli Shechtman, Oliver Wang, and Hao Li
Style Transfer (Harmonization)
A Neural Algorithm of Artistic Style, 2015, Leon A. Gatys, Alexander S. Ecker,
Matthias Bethge
Copyright © 2018 by AI4quant [TW] All rights reserved
Proposed model#1 - PConv
Hole mask: random shape
Copyright © 2018 by AI4quant [TW] All rights reserved
U-net like architecture(encoder-decoder)
Copyright © 2018 by AI4quant [TW] All rights reserved
PConv:During prediction-Erosion
Copyright © 2018 by AI4quant [TW] All rights reserved
Proposed model#2: Content network- central square
Model : pre-trained imagenet_inpaintCenter.t7
Copyright © 2018 by AI4quant [TW] All rights reserved
Final image: Image blending
g(x)=(1-ß)*f1(x)+ß*f2(x)
ß is (0,1)
Copyright © 2018 by AI4quant [TW] All rights reserved
Code library and computation platform
● Tensorflow 1.9.0
● Keras 2.2.2
● OpenCV 3.4.0
● Torch
Training(Oregon, U.S.) Model Evaluation/Competition (GCP in Taiwan)
● AWS p3.2xlarge
● Nvidia V100 GPU
● 8vCPU
● 61GB memory
● GCP Google Compute Engine
● Nvidia K80 GPU
● 4 vCPU
● 26 GB memory
Copyright © 2018 by AI4quant [TW] All rights reserved
Quantitative comparison(validation data loss)
Copyright © 2018 by AI4quant [TW] All rights reserved
Qualitative comparison
Copyright © 2018 by AI4quant [TW] All rights reserved
During prediction - Deep style transfer(harmonization)
(Thinking to use)
Iter 0 Iter 5 Iter 10

Weitere ähnliche Inhalte

Ähnlich wie AI4quant 2018 PIXNET Hackathon

Quick! Quick! Exploration!: A framework for searching a predictive model on A...
Quick! Quick! Exploration!: A framework for searching a predictive model on A...Quick! Quick! Exploration!: A framework for searching a predictive model on A...
Quick! Quick! Exploration!: A framework for searching a predictive model on A...
DataWorks Summit
 
Fri benghiat gil-odsc-data-kitchen-data science to dataops
Fri benghiat gil-odsc-data-kitchen-data science to dataopsFri benghiat gil-odsc-data-kitchen-data science to dataops
Fri benghiat gil-odsc-data-kitchen-data science to dataops
DataKitchen
 

Ähnlich wie AI4quant 2018 PIXNET Hackathon (20)

OpenPOWER Boot camp in Zurich
OpenPOWER Boot camp in ZurichOpenPOWER Boot camp in Zurich
OpenPOWER Boot camp in Zurich
 
Lossless Encryption using BITPLANE and EDGEMAP Crypt Algorithms
Lossless Encryption using BITPLANE and EDGEMAP Crypt AlgorithmsLossless Encryption using BITPLANE and EDGEMAP Crypt Algorithms
Lossless Encryption using BITPLANE and EDGEMAP Crypt Algorithms
 
All AI Roads lead to Distribution - Dot AI
All AI Roads lead to Distribution - Dot AIAll AI Roads lead to Distribution - Dot AI
All AI Roads lead to Distribution - Dot AI
 
lec1b.ppt
lec1b.pptlec1b.ppt
lec1b.ppt
 
Distributed TensorFlow on Hops (Papis London, April 2018)
Distributed TensorFlow on Hops (Papis London, April 2018)Distributed TensorFlow on Hops (Papis London, April 2018)
Distributed TensorFlow on Hops (Papis London, April 2018)
 
Jfokus 2019-dowling-logical-clocks
Jfokus 2019-dowling-logical-clocksJfokus 2019-dowling-logical-clocks
Jfokus 2019-dowling-logical-clocks
 
Gettingstartedwithmatlabimageprocessing
GettingstartedwithmatlabimageprocessingGettingstartedwithmatlabimageprocessing
Gettingstartedwithmatlabimageprocessing
 
Quick! Quick! Exploration!: A framework for searching a predictive model on A...
Quick! Quick! Exploration!: A framework for searching a predictive model on A...Quick! Quick! Exploration!: A framework for searching a predictive model on A...
Quick! Quick! Exploration!: A framework for searching a predictive model on A...
 
Arcelormittal @ Scilab Conference 2018
Arcelormittal @ Scilab Conference 2018Arcelormittal @ Scilab Conference 2018
Arcelormittal @ Scilab Conference 2018
 
ODSC18, London, How to build high performing weighted XGBoost ML Model for Re...
ODSC18, London, How to build high performing weighted XGBoost ML Model for Re...ODSC18, London, How to build high performing weighted XGBoost ML Model for Re...
ODSC18, London, How to build high performing weighted XGBoost ML Model for Re...
 
How to Leverage Machine Learning (R, Hadoop, Spark, H2O) for Real Time Proces...
How to Leverage Machine Learning (R, Hadoop, Spark, H2O) for Real Time Proces...How to Leverage Machine Learning (R, Hadoop, Spark, H2O) for Real Time Proces...
How to Leverage Machine Learning (R, Hadoop, Spark, H2O) for Real Time Proces...
 
Pitch v2.2
Pitch v2.2Pitch v2.2
Pitch v2.2
 
Webinar: Deep Learning Pipelines Beyond the Learning
Webinar: Deep Learning Pipelines Beyond the LearningWebinar: Deep Learning Pipelines Beyond the Learning
Webinar: Deep Learning Pipelines Beyond the Learning
 
Meg Mude, Intel - Data Engineering Lifecycle Optimized on Intel - H2O World S...
Meg Mude, Intel - Data Engineering Lifecycle Optimized on Intel - H2O World S...Meg Mude, Intel - Data Engineering Lifecycle Optimized on Intel - H2O World S...
Meg Mude, Intel - Data Engineering Lifecycle Optimized on Intel - H2O World S...
 
Incorporating JanusGraph into your Scylla Ecosystem
Incorporating JanusGraph into your Scylla EcosystemIncorporating JanusGraph into your Scylla Ecosystem
Incorporating JanusGraph into your Scylla Ecosystem
 
One bridge to connect them all. Oracle GoldenGate for Big Data.UKOUG Tech 2018
One bridge to connect them all. Oracle GoldenGate for Big Data.UKOUG Tech 2018One bridge to connect them all. Oracle GoldenGate for Big Data.UKOUG Tech 2018
One bridge to connect them all. Oracle GoldenGate for Big Data.UKOUG Tech 2018
 
Fri benghiat gil-odsc-data-kitchen-data science to dataops
Fri benghiat gil-odsc-data-kitchen-data science to dataopsFri benghiat gil-odsc-data-kitchen-data science to dataops
Fri benghiat gil-odsc-data-kitchen-data science to dataops
 
ODSC data science to DataOps
ODSC data science to DataOpsODSC data science to DataOps
ODSC data science to DataOps
 
Northwestern 20181004 v9
Northwestern 20181004 v9Northwestern 20181004 v9
Northwestern 20181004 v9
 
"Approaches for Energy Efficient Implementation of Deep Neural Networks," a P...
"Approaches for Energy Efficient Implementation of Deep Neural Networks," a P..."Approaches for Energy Efficient Implementation of Deep Neural Networks," a P...
"Approaches for Energy Efficient Implementation of Deep Neural Networks," a P...
 

Mehr von Tzu-Cheng(Jason) Chuang

Mehr von Tzu-Cheng(Jason) Chuang (6)

Report Cruncher
Report CruncherReport Cruncher
Report Cruncher
 
昕力技術小聚2023-02-23 OAuth 2.0 Spring Boot 3.0
昕力技術小聚2023-02-23 OAuth 2.0 Spring Boot 3.0昕力技術小聚2023-02-23 OAuth 2.0 Spring Boot 3.0
昕力技術小聚2023-02-23 OAuth 2.0 Spring Boot 3.0
 
AI model in smart manufacturer
AI model in smart manufacturerAI model in smart manufacturer
AI model in smart manufacturer
 
AI4quant decks in AppWorks DemoDay #17
AI4quant decks in AppWorks DemoDay #17AI4quant decks in AppWorks DemoDay #17
AI4quant decks in AppWorks DemoDay #17
 
ai4quant TMU MIT Hackathon 2018
ai4quant TMU MIT Hackathon 2018ai4quant TMU MIT Hackathon 2018
ai4quant TMU MIT Hackathon 2018
 
AIQUAN platform for AIoT
AIQUAN platform for AIoTAIQUAN platform for AIoT
AIQUAN platform for AIoT
 

Kürzlich hochgeladen

+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
Health
 
Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024
Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024
Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024
VictoriaMetrics
 
%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...
%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...
%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...
masabamasaba
 
Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...
Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...
Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...
Medical / Health Care (+971588192166) Mifepristone and Misoprostol tablets 200mg
 
%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...
masabamasaba
 
%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...
masabamasaba
 
%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...
%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...
%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...
masabamasaba
 

Kürzlich hochgeladen (20)

%in Rustenburg+277-882-255-28 abortion pills for sale in Rustenburg
%in Rustenburg+277-882-255-28 abortion pills for sale in Rustenburg%in Rustenburg+277-882-255-28 abortion pills for sale in Rustenburg
%in Rustenburg+277-882-255-28 abortion pills for sale in Rustenburg
 
AI & Machine Learning Presentation Template
AI & Machine Learning Presentation TemplateAI & Machine Learning Presentation Template
AI & Machine Learning Presentation Template
 
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
+971565801893>>SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHAB...
 
OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...
OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...
OpenChain - The Ramifications of ISO/IEC 5230 and ISO/IEC 18974 for Legal Pro...
 
Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024
Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024
Large-scale Logging Made Easy: Meetup at Deutsche Bank 2024
 
%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...
%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...
%+27788225528 love spells in Knoxville Psychic Readings, Attraction spells,Br...
 
%in tembisa+277-882-255-28 abortion pills for sale in tembisa
%in tembisa+277-882-255-28 abortion pills for sale in tembisa%in tembisa+277-882-255-28 abortion pills for sale in tembisa
%in tembisa+277-882-255-28 abortion pills for sale in tembisa
 
Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...
Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...
Abortion Pills In Pretoria ](+27832195400*)[ 🏥 Women's Abortion Clinic In Pre...
 
Microsoft AI Transformation Partner Playbook.pdf
Microsoft AI Transformation Partner Playbook.pdfMicrosoft AI Transformation Partner Playbook.pdf
Microsoft AI Transformation Partner Playbook.pdf
 
%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Toronto Psychic Readings, Attraction spells,Brin...
 
%in Harare+277-882-255-28 abortion pills for sale in Harare
%in Harare+277-882-255-28 abortion pills for sale in Harare%in Harare+277-882-255-28 abortion pills for sale in Harare
%in Harare+277-882-255-28 abortion pills for sale in Harare
 
%in Soweto+277-882-255-28 abortion pills for sale in soweto
%in Soweto+277-882-255-28 abortion pills for sale in soweto%in Soweto+277-882-255-28 abortion pills for sale in soweto
%in Soweto+277-882-255-28 abortion pills for sale in soweto
 
WSO2CON2024 - It's time to go Platformless
WSO2CON2024 - It's time to go PlatformlessWSO2CON2024 - It's time to go Platformless
WSO2CON2024 - It's time to go Platformless
 
%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...
%+27788225528 love spells in Atlanta Psychic Readings, Attraction spells,Brin...
 
%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...
%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...
%+27788225528 love spells in Boston Psychic Readings, Attraction spells,Bring...
 
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
 
Announcing Codolex 2.0 from GDK Software
Announcing Codolex 2.0 from GDK SoftwareAnnouncing Codolex 2.0 from GDK Software
Announcing Codolex 2.0 from GDK Software
 
Direct Style Effect Systems - The Print[A] Example - A Comprehension Aid
Direct Style Effect Systems -The Print[A] Example- A Comprehension AidDirect Style Effect Systems -The Print[A] Example- A Comprehension Aid
Direct Style Effect Systems - The Print[A] Example - A Comprehension Aid
 
Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...
Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...
Shapes for Sharing between Graph Data Spaces - and Epistemic Querying of RDF-...
 
%in kaalfontein+277-882-255-28 abortion pills for sale in kaalfontein
%in kaalfontein+277-882-255-28 abortion pills for sale in kaalfontein%in kaalfontein+277-882-255-28 abortion pills for sale in kaalfontein
%in kaalfontein+277-882-255-28 abortion pills for sale in kaalfontein
 

AI4quant 2018 PIXNET Hackathon

  • 1. Copyright © 2018 by AI4quant [TW] All rights reserved AI4quant: AI智慧影像生成 PIXNET Hackathon 2018 Jason Chuang、呂沂樺、Kiki 2018-08-19
  • 2. Copyright © 2018 by AI4quant [TW] All rights reserved What is AI? Big Data Algorithm Computing Power
  • 3. Copyright © 2018 by AI4quant [TW] All rights reserved Data preprocessing for training data We use 5 different datasets for training If the shortest width of the image is less than 256 pixels, ignore Shrink the shortest width to 256 pixels, cut the center part of the image
  • 4. Copyright © 2018 by AI4quant [TW] All rights reserved Big data Food-11: https://mmspg.epfl.ch/food-image-datasets DataSet Size(jpg, png) Number of Files # Training files Training size(bmp, 256x256) # Validation files dataset100 1GB 14,467 10,044 1.83GB 0 dataset256 4GB 31,653 27,002 5GB 0 Food-11 1.1GB 16,643 16,635 3.04GB 0 food-101 4.76GB 101,008 100,956 18.4GB 0 pixfood20 2.03GB 31,305 19,322 3.53GB 19,322
  • 5. Copyright © 2018 by AI4quant [TW] All rights reserved Algorithms are motivated by research paper DCNN based method Image Inpainting for Irregular Holes Using Partial Convolutions, 2018, Guilin Liu, Fitsum A. Reda, Kevin J. Shih, Ting-Chun Wang, Andrew Tao, Bryan Catanzaro High-Resolution Image Inpainting using Multi-Scale Neural Patch Synthesis, 2017, Chao Yang, Xin Lu,Zhe Lin, Eli Shechtman, Oliver Wang, and Hao Li Style Transfer (Harmonization) A Neural Algorithm of Artistic Style, 2015, Leon A. Gatys, Alexander S. Ecker, Matthias Bethge
  • 6. Copyright © 2018 by AI4quant [TW] All rights reserved Proposed model#1 - PConv Hole mask: random shape
  • 7. Copyright © 2018 by AI4quant [TW] All rights reserved U-net like architecture(encoder-decoder)
  • 8. Copyright © 2018 by AI4quant [TW] All rights reserved PConv:During prediction-Erosion
  • 9. Copyright © 2018 by AI4quant [TW] All rights reserved Proposed model#2: Content network- central square Model : pre-trained imagenet_inpaintCenter.t7
  • 10. Copyright © 2018 by AI4quant [TW] All rights reserved Final image: Image blending g(x)=(1-ß)*f1(x)+ß*f2(x) ß is (0,1)
  • 11. Copyright © 2018 by AI4quant [TW] All rights reserved Code library and computation platform ● Tensorflow 1.9.0 ● Keras 2.2.2 ● OpenCV 3.4.0 ● Torch Training(Oregon, U.S.) Model Evaluation/Competition (GCP in Taiwan) ● AWS p3.2xlarge ● Nvidia V100 GPU ● 8vCPU ● 61GB memory ● GCP Google Compute Engine ● Nvidia K80 GPU ● 4 vCPU ● 26 GB memory
  • 12. Copyright © 2018 by AI4quant [TW] All rights reserved Quantitative comparison(validation data loss)
  • 13. Copyright © 2018 by AI4quant [TW] All rights reserved Qualitative comparison
  • 14. Copyright © 2018 by AI4quant [TW] All rights reserved During prediction - Deep style transfer(harmonization) (Thinking to use) Iter 0 Iter 5 Iter 10