SlideShare a Scribd company logo
Image-to-image translation with
Generative Adversarial Networks
(without math)
Gabriele Graffieti
Research Fellow
Alma Mater Studiorum Universit`a di Bologna
gabriele.graffieti@unibo.it
May 31, 2019
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 1 / 25
Image-to-image translation I
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 2 / 25
Image-to-image translation II
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 3 / 25
Image-to-image translation III
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 4 / 25
Image-to-image translation IV
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 5 / 25
Image-to-image translation V
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 6 / 25
Image-to-image translation VI
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 7 / 25
Image-to-image translation VII
How do we do it?
• We know features of the original domain that should not be present
in the results (snow, fog).
• We know features of the target domain that should be added in the
results (blue sky, hallucinated colors).
Nobody teach us what a summer scene looks like, we learn it from
data.
Key concept is realism!
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 8 / 25
Image-to-image translation VIII
And machines?
• The can learn concepts directly from data (machine learning).
• But how can we evaluate realism of the results?
• We are not able to define a metric for summerness or van goghness.
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 9 / 25
Image-to-image translation IX
My beloved opponent
• Translation can be seen as a particular case of imitation.
• The translated image should be similar to those on the target domain
(realism).
• We should exploit the forger-police officer game.
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 10 / 25
Proto-GAN framework
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 11 / 25
GAN framework for image-to-image translation
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 12 / 25
But the image content?
How we can ensure that the image content remain the same?
• The adversarial loss only ensure that the translated image looks
similar to the ones in the target domain.
• The content can be changed without any penalization.
• We don’t have a reference image in the target domain to compare
with the translated image.
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 13 / 25
Cycle consistency
Given a mapping G : X → Y and its inverse F : Y → X the two mappings
should be cycle consistent with each other.
F(G(x)) ≈ x
and
G(F(y)) ≈ y
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 14 / 25
CycleGAN
The CycleGAN model is composed of two different GANs, which learn
inverse translations between two image domains. These translations are
maintained cycle consistent through the difference between original and
recovered images.
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 15 / 25
Neural style transfer art I
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 16 / 25
Neural style transfer art II
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 17 / 25
Neural style transfer art III
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 18 / 25
Neural style transfer art IV
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 19 / 25
Photoshop 2.0?
• https:
//www.youtube.com/watch?time_continue=25&v=MXWm6w4E5q0
• http://gandissect.res.ibm.com/ganpaint.html?project=
churchoutdoor&layer=layer4
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 20 / 25
Deepfakes I
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 21 / 25
Deepfakes II
Good Utilizations
• Special effects in movies.
• ??
(Very) Bad Uses
• Fake porn.
• Video forging (discredit a politician, direct popular opinions, . . . ).
• False evidences in court cases.
• . . .
We will be able to distinguish fake data from reality?
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 22 / 25
Defogging I
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 23 / 25
Defogging II
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 24 / 25
Image-to-image translation with
Generative Adversarial Networks
(without math)
Gabriele Graffieti
Research Fellow
Alma Mater Studiorum Universit`a di Bologna
gabriele.graffieti@unibo.it
May 31, 2019
Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 25 / 25

More Related Content

Recently uploaded

Search and Society: Reimagining Information Access for Radical Futures
Search and Society: Reimagining Information Access for Radical FuturesSearch and Society: Reimagining Information Access for Radical Futures
Search and Society: Reimagining Information Access for Radical Futures
Bhaskar Mitra
 

Recently uploaded (20)

GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
 
From Daily Decisions to Bottom Line: Connecting Product Work to Revenue by VP...
From Daily Decisions to Bottom Line: Connecting Product Work to Revenue by VP...From Daily Decisions to Bottom Line: Connecting Product Work to Revenue by VP...
From Daily Decisions to Bottom Line: Connecting Product Work to Revenue by VP...
 
Search and Society: Reimagining Information Access for Radical Futures
Search and Society: Reimagining Information Access for Radical FuturesSearch and Society: Reimagining Information Access for Radical Futures
Search and Society: Reimagining Information Access for Radical Futures
 
Speed Wins: From Kafka to APIs in Minutes
Speed Wins: From Kafka to APIs in MinutesSpeed Wins: From Kafka to APIs in Minutes
Speed Wins: From Kafka to APIs in Minutes
 
Exploring UiPath Orchestrator API: updates and limits in 2024 🚀
Exploring UiPath Orchestrator API: updates and limits in 2024 🚀Exploring UiPath Orchestrator API: updates and limits in 2024 🚀
Exploring UiPath Orchestrator API: updates and limits in 2024 🚀
 
10 Differences between Sales Cloud and CPQ, Blanka Doktorová
10 Differences between Sales Cloud and CPQ, Blanka Doktorová10 Differences between Sales Cloud and CPQ, Blanka Doktorová
10 Differences between Sales Cloud and CPQ, Blanka Doktorová
 
Empowering NextGen Mobility via Large Action Model Infrastructure (LAMI): pav...
Empowering NextGen Mobility via Large Action Model Infrastructure (LAMI): pav...Empowering NextGen Mobility via Large Action Model Infrastructure (LAMI): pav...
Empowering NextGen Mobility via Large Action Model Infrastructure (LAMI): pav...
 
UiPath Test Automation using UiPath Test Suite series, part 3
UiPath Test Automation using UiPath Test Suite series, part 3UiPath Test Automation using UiPath Test Suite series, part 3
UiPath Test Automation using UiPath Test Suite series, part 3
 
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered QualitySoftware Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
 
Mission to Decommission: Importance of Decommissioning Products to Increase E...
Mission to Decommission: Importance of Decommissioning Products to Increase E...Mission to Decommission: Importance of Decommissioning Products to Increase E...
Mission to Decommission: Importance of Decommissioning Products to Increase E...
 
SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...
SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...
SOQL 201 for Admins & Developers: Slice & Dice Your Org’s Data With Aggregate...
 
Designing Great Products: The Power of Design and Leadership by Chief Designe...
Designing Great Products: The Power of Design and Leadership by Chief Designe...Designing Great Products: The Power of Design and Leadership by Chief Designe...
Designing Great Products: The Power of Design and Leadership by Chief Designe...
 
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
 
Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)
Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)
Measures in SQL (a talk at SF Distributed Systems meetup, 2024-05-22)
 
Integrating Telephony Systems with Salesforce: Insights and Considerations, B...
Integrating Telephony Systems with Salesforce: Insights and Considerations, B...Integrating Telephony Systems with Salesforce: Insights and Considerations, B...
Integrating Telephony Systems with Salesforce: Insights and Considerations, B...
 
How world-class product teams are winning in the AI era by CEO and Founder, P...
How world-class product teams are winning in the AI era by CEO and Founder, P...How world-class product teams are winning in the AI era by CEO and Founder, P...
How world-class product teams are winning in the AI era by CEO and Founder, P...
 
To Graph or Not to Graph Knowledge Graph Architectures and LLMs
To Graph or Not to Graph Knowledge Graph Architectures and LLMsTo Graph or Not to Graph Knowledge Graph Architectures and LLMs
To Graph or Not to Graph Knowledge Graph Architectures and LLMs
 
From Siloed Products to Connected Ecosystem: Building a Sustainable and Scala...
From Siloed Products to Connected Ecosystem: Building a Sustainable and Scala...From Siloed Products to Connected Ecosystem: Building a Sustainable and Scala...
From Siloed Products to Connected Ecosystem: Building a Sustainable and Scala...
 
AI revolution and Salesforce, Jiří Karpíšek
AI revolution and Salesforce, Jiří KarpíšekAI revolution and Salesforce, Jiří Karpíšek
AI revolution and Salesforce, Jiří Karpíšek
 
Free and Effective: Making Flows Publicly Accessible, Yumi Ibrahimzade
Free and Effective: Making Flows Publicly Accessible, Yumi IbrahimzadeFree and Effective: Making Flows Publicly Accessible, Yumi Ibrahimzade
Free and Effective: Making Flows Publicly Accessible, Yumi Ibrahimzade
 

Featured

How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental Health
ThinkNow
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie Insights
Kurio // The Social Media Age(ncy)
 

Featured (20)

2024 State of Marketing Report – by Hubspot
2024 State of Marketing Report – by Hubspot2024 State of Marketing Report – by Hubspot
2024 State of Marketing Report – by Hubspot
 
Everything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTEverything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPT
 
Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage Engineerings
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental Health
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
 
Skeleton Culture Code
Skeleton Culture CodeSkeleton Culture Code
Skeleton Culture Code
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie Insights
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search Intent
 
How to have difficult conversations
How to have difficult conversations How to have difficult conversations
How to have difficult conversations
 
Introduction to Data Science
Introduction to Data ScienceIntroduction to Data Science
Introduction to Data Science
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best Practices
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project management
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
 

Image-to-image Translation with Generative Adversarial Networks (without math)

  • 1. Image-to-image translation with Generative Adversarial Networks (without math) Gabriele Graffieti Research Fellow Alma Mater Studiorum Universit`a di Bologna gabriele.graffieti@unibo.it May 31, 2019 Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 1 / 25
  • 2. Image-to-image translation I Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 2 / 25
  • 3. Image-to-image translation II Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 3 / 25
  • 4. Image-to-image translation III Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 4 / 25
  • 5. Image-to-image translation IV Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 5 / 25
  • 6. Image-to-image translation V Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 6 / 25
  • 7. Image-to-image translation VI Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 7 / 25
  • 8. Image-to-image translation VII How do we do it? • We know features of the original domain that should not be present in the results (snow, fog). • We know features of the target domain that should be added in the results (blue sky, hallucinated colors). Nobody teach us what a summer scene looks like, we learn it from data. Key concept is realism! Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 8 / 25
  • 9. Image-to-image translation VIII And machines? • The can learn concepts directly from data (machine learning). • But how can we evaluate realism of the results? • We are not able to define a metric for summerness or van goghness. Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 9 / 25
  • 10. Image-to-image translation IX My beloved opponent • Translation can be seen as a particular case of imitation. • The translated image should be similar to those on the target domain (realism). • We should exploit the forger-police officer game. Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 10 / 25
  • 11. Proto-GAN framework Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 11 / 25
  • 12. GAN framework for image-to-image translation Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 12 / 25
  • 13. But the image content? How we can ensure that the image content remain the same? • The adversarial loss only ensure that the translated image looks similar to the ones in the target domain. • The content can be changed without any penalization. • We don’t have a reference image in the target domain to compare with the translated image. Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 13 / 25
  • 14. Cycle consistency Given a mapping G : X → Y and its inverse F : Y → X the two mappings should be cycle consistent with each other. F(G(x)) ≈ x and G(F(y)) ≈ y Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 14 / 25
  • 15. CycleGAN The CycleGAN model is composed of two different GANs, which learn inverse translations between two image domains. These translations are maintained cycle consistent through the difference between original and recovered images. Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 15 / 25
  • 16. Neural style transfer art I Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 16 / 25
  • 17. Neural style transfer art II Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 17 / 25
  • 18. Neural style transfer art III Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 18 / 25
  • 19. Neural style transfer art IV Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 19 / 25
  • 20. Photoshop 2.0? • https: //www.youtube.com/watch?time_continue=25&v=MXWm6w4E5q0 • http://gandissect.res.ibm.com/ganpaint.html?project= churchoutdoor&layer=layer4 Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 20 / 25
  • 21. Deepfakes I Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 21 / 25
  • 22. Deepfakes II Good Utilizations • Special effects in movies. • ?? (Very) Bad Uses • Fake porn. • Video forging (discredit a politician, direct popular opinions, . . . ). • False evidences in court cases. • . . . We will be able to distinguish fake data from reality? Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 22 / 25
  • 23. Defogging I Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 23 / 25
  • 24. Defogging II Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 24 / 25
  • 25. Image-to-image translation with Generative Adversarial Networks (without math) Gabriele Graffieti Research Fellow Alma Mater Studiorum Universit`a di Bologna gabriele.graffieti@unibo.it May 31, 2019 Gabriele Graffieti Image-to-image translation with GANs May 31, 2019 25 / 25