SlideShare a Scribd company logo
AI 的逆襲,
眼⾒見見真的為憑嗎?
台灣資安⼤大會 iThome CYBERSEC 2019

Turkey & Melody  

台灣駭客協會 秘書長
HITCON GIRLS 共同創辦⼈人
台灣駭客協會 專案經理理
HITCON GIRLS 成員
ABOUT US
2
Turkey Melody
OUTLINE
3
• Deepfakes
• Lyrebirds
• 應⽤用
• 反制技術
DEEPFAKES MAKES IT POSSIBLE TO
CREATE FAKE VIDEOS OF ALMOST
EVERYONE.
4
5
LET’S TALK ABOUT DEEPFAKES!
1. In December 2017, a user
named “DeepFakes”
announce the tool on
Reddit Community of
developers.
2. DeepFakes is a tool that
utilizes deep learning to
recognize and swap faces
in pictures and videos.
5
6
HOW DEEPFAKES WORK?
HOW DEEPFAKES WORK?
https://github.com/iperov/DeepFaceLab
7
HOW DEEPFAKES WORK?
>DeepFakes Tools
1. Fakeapp
2. DeepFaceLab
3. Fakeswap
4. Openfaceswap
5. Myfakeapp
8
HOW DEEPFAKES WORK?
>DeepFakes Tools
1. Fakeapp
2.DeepFaceLab
3. Fakeswap
4. Openfaceswap
5. Myfakeapp
>
9
Source Target
10
HOW DEEPFAKES WORK?
>Steps for: DeepFaceLab
2 31
Extract image
from source
video.
Extract faces
from source
and target
image.
4
Manually
remove
error image.
5
Model
Training
6
Extract image
from target
video.
Debug & Convert
image to mp4
11
HOW DEEPFAKES WORK?
>Extract image from source&target video.
FFmpeg
FFmpeg is the leading
multimedia framework,
able to decode, encode,
transcode, mux, demux,
stream, filter and play
pretty much anything
that humans and
machines have created.
ffmpeg -i clipname -vf fps=framerate -qscale:v 2 "imagename%04d.jpg"Command_
12
HOW DEEPFAKES WORK?
>Extract faces from source and target image
(Ref: Facial landmarks with dlib, OpenCV, and Python)
Divided a face into the following areas:
1. eyes(left/right)
2. eyebrows(left/right)
3. nose
4. chin
13
HOW DEEPFAKES WORK?
>Extract faces from source and target image
(Ref: Real-Time Face Pose Estimation )
14
HOW DEEPFAKES WORK?
>Extract faces from source and target image
15
HOW DEEPFAKES WORK?
>Extract faces from source and target image
(Ref: One Millisecond Face Alignment with an Ensemble of Regression Trees)16
HOW DEEPFAKES WORK?
>Manually remove error image
17
HOW DEEPFAKES WORK?
>Manually remove error image
18
HOW DEEPFAKES WORK?
>Manually remove error image
19
HOW DEEPFAKES WORK?
>Model Training
(Ref: https://deepfakes.com.cn/)20
HOW DEEPFAKES WORK?
>Model Training
(Ref: https://deepfakes.com.cn/)21
HOW DEEPFAKES WORK?
>Model Training
(Ref: https://deepfakes.com.cn/)22
HOW DEEPFAKES WORK?
>Model Training
Models Types:
1. H64
2. H128
3. DF
4. LIAEF
5. LIAEF128YAW
6. MIAEF128
7. AVATAR
8. SAE H12823
HOW DEEPFAKES WORK?
>Model Training
Models Types:
1. H64
2. H128
3. DF
4. LIAEF
5. LIAEF128YAW
6. MIAEF128
7. AVATAR
8. SAE H128
Train 越久效果越好!
24
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
25
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
26
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
27
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
28
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
29
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
30
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
Debug Mode
Use predicted mask? 1
Erosion (-100 to +100): (default = 0) 1
Seamless Erosion (0 to 40): (default = 0) 20
Blur (-200 to +200): (default = 0) 40
Hist-match threshold (0 to 255): (default = 255) default
Face Scale (-50 to +50): (default = 0) default
Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT
Degrade Color Power of Final Image: (default = 0) default
31
HOW DEEPFAKES WORK?
>Debug & Convert image to mp4
ffmpeg -f image2 -i imagename%04d.jpg -vcodec libx264 -crf 15 -pix_fmt yuv420pCommand_
32
(Downlink: https://github.com/iperov/DeepFaceLab)
HOW DEEPFAKES WORK?
>Download DeepFaceLab
33
34
LET’S DEMO !
HOW DEEPFAKES WORK?
>⼩小插曲
實測跑評:RTX 2080 ⼤大約需跑 6 個⼩小時以上會有不錯的成果
作者跪求 RTX
2080 Ti 中 XD
35
HOW DEEPFAKES WORK?
>不要讓貧窮限制你的 AI
#讓 AI 聰明前
先飽滿你的⼝口袋
36
Source Target
37
38
(Link:https://youtu.be/yfsdZ5gLuOo)
CAN AI MIMIC YOUR VOICE?
39
CAN YOU TELL A FAKE VOICE FROM A REAL ONE?
40
(Ref: https://lyrebird.ai/)
• create a digitized version of your own voice.
• Uses deep learning frameworks on AWS to develop and
train its AI models
What is Lyrebird?
Record & Upload
voice samples
Train by DL Framework
on AWS EC2 P3
Digital Voice
Generated
Optput audio of
requested dialog
(Ref: https://lyrebird.ai/)41
LYREBIRD LIVE DEMO
This is my honored to talk to everyone here.
However, can you trust your eyes and ears to perceive reality?
Machine learning can artificially mimic natural sounds to create digital voice.
Just like me!
42
Why my voice sounds awful?
43
(Ref: https://lyrebird.ai/)
Voice Cloning Market
Generation Is Coming
44
CHALLENGES FOR OUR FUTURE
45
Why do you believe what
I am telling you now?
NEGATIVE EFFECTS
46
POSITIVE EFFECTS
Create images and voices
from the past.
47
(Ref:apple)
(Ref:Fast & Furious 6)
(Ref:https://www.youtube.com/watch?v=5iZuffHPDAw)
FIGHT AGAINST FAKE VIDEOS!
49
FAKE VIDEO DETECT SOLUTION
>Detect malicious alterations with AI
•abnormal compression signatures
•lip sync analysis
•Video metadata analysis
•noise patterns analysis
50
FAKE VIDEO DETECT SOLUTION
> Fingerprint source videos & track
provenance with blockchain smart contracts
Create Hashing Blackchan record submitt
51
REFERENCE
• https://deepfakes.com.cn/
• https://www.youtube.com/watch?v=K98nTNjXkq8
• ttps://github.com/Fabsqrt/BitTigerLab
• https://github.com/iperov/DeepFaceLab
• https://www.youtube.com/watch?v=CR5Jr6Z1KI
• https://lyrebird.ai/
• https://www.darpa.mil/
• https://medium.com/amber-video
52
THANKS!
53

More Related Content

Similar to iThome 資安大會 2019 資安大會_AI 的逆襲,眼見真的為憑嗎? - Turkey&Melody.pdf

Fastest css3 animations
Fastest css3 animations Fastest css3 animations
Fastest css3 animations
Ismail Berkay Beyaz
 
클라우드 네트워킹과 SDN 그리고 OpenStack
클라우드 네트워킹과 SDN 그리고 OpenStack클라우드 네트워킹과 SDN 그리고 OpenStack
클라우드 네트워킹과 SDN 그리고 OpenStack
Choe Cheng-Dae
 
Key codes
Key codesKey codes
Key codes
tusher khan
 
HTML5, CSS3, and other fancy buzzwords
HTML5, CSS3, and other fancy buzzwordsHTML5, CSS3, and other fancy buzzwords
HTML5, CSS3, and other fancy buzzwords
Mo Jangda
 
C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...
C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...
C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...
Amazon Web Services
 
Flash Is Dead, Long Live Flash!
Flash Is Dead, Long Live Flash!Flash Is Dead, Long Live Flash!
Flash Is Dead, Long Live Flash!
Rye Corradini
 
Triton and Symbolic execution on GDB@DEF CON China
Triton and Symbolic execution on GDB@DEF CON ChinaTriton and Symbolic execution on GDB@DEF CON China
Triton and Symbolic execution on GDB@DEF CON China
Wei-Bo Chen
 
Hacking With Pictures SyScan 2015
Hacking With Pictures SyScan 2015Hacking With Pictures SyScan 2015
Hacking With Pictures SyScan 2015
Saumil Shah
 
Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...
Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...
Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...
confluent
 
Image Processing Using MATLAB
Image Processing Using MATLABImage Processing Using MATLAB
Image Processing Using MATLAB
Amarjeetsingh Thakur
 
Image processing in MATLAB
Image processing in MATLABImage processing in MATLAB
Image processing in MATLAB
Amarjeetsingh Thakur
 
All your CSS are belong to us
All your CSS are belong to usAll your CSS are belong to us
All your CSS are belong to us
michalbu
 
Kanban for Software Development and Kaizen Culture
Kanban for Software Development and Kaizen CultureKanban for Software Development and Kaizen Culture
Kanban for Software Development and Kaizen Culture
Acquate
 
Web Presentations, deck.js and Extensions
Web Presentations, deck.js and ExtensionsWeb Presentations, deck.js and Extensions
Web Presentations, deck.js and Extensions
remiemonet
 
[Code night] natural language proccessing and machine learning
[Code night] natural language proccessing and machine learning[Code night] natural language proccessing and machine learning
[Code night] natural language proccessing and machine learning
Kenichi Sonoda
 
孫民/從電腦視覺看人工智慧 : 下一件大事
孫民/從電腦視覺看人工智慧 : 下一件大事孫民/從電腦視覺看人工智慧 : 下一件大事
孫民/從電腦視覺看人工智慧 : 下一件大事
台灣資料科學年會
 

Similar to iThome 資安大會 2019 資安大會_AI 的逆襲,眼見真的為憑嗎? - Turkey&Melody.pdf (20)

Fastest css3 animations
Fastest css3 animations Fastest css3 animations
Fastest css3 animations
 
클라우드 네트워킹과 SDN 그리고 OpenStack
클라우드 네트워킹과 SDN 그리고 OpenStack클라우드 네트워킹과 SDN 그리고 OpenStack
클라우드 네트워킹과 SDN 그리고 OpenStack
 
Key codes
Key codesKey codes
Key codes
 
HTML5, CSS3, and other fancy buzzwords
HTML5, CSS3, and other fancy buzzwordsHTML5, CSS3, and other fancy buzzwords
HTML5, CSS3, and other fancy buzzwords
 
C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...
C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...
C5 Instances and the Evolution of Amazon EC2 Virtualization - CMP332 - re:Inv...
 
Flash Is Dead, Long Live Flash!
Flash Is Dead, Long Live Flash!Flash Is Dead, Long Live Flash!
Flash Is Dead, Long Live Flash!
 
Triton and Symbolic execution on GDB@DEF CON China
Triton and Symbolic execution on GDB@DEF CON ChinaTriton and Symbolic execution on GDB@DEF CON China
Triton and Symbolic execution on GDB@DEF CON China
 
Hacking With Pictures SyScan 2015
Hacking With Pictures SyScan 2015Hacking With Pictures SyScan 2015
Hacking With Pictures SyScan 2015
 
Image filters
Image filtersImage filters
Image filters
 
Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...
Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...
Streaming Apps and Poison Pills: handle the unexpected with Kafka Streams (Lo...
 
Css3 process bar
Css3 process barCss3 process bar
Css3 process bar
 
Css3 process bar
Css3 process barCss3 process bar
Css3 process bar
 
Css3 process bar
Css3 process barCss3 process bar
Css3 process bar
 
Image Processing Using MATLAB
Image Processing Using MATLABImage Processing Using MATLAB
Image Processing Using MATLAB
 
Image processing in MATLAB
Image processing in MATLABImage processing in MATLAB
Image processing in MATLAB
 
All your CSS are belong to us
All your CSS are belong to usAll your CSS are belong to us
All your CSS are belong to us
 
Kanban for Software Development and Kaizen Culture
Kanban for Software Development and Kaizen CultureKanban for Software Development and Kaizen Culture
Kanban for Software Development and Kaizen Culture
 
Web Presentations, deck.js and Extensions
Web Presentations, deck.js and ExtensionsWeb Presentations, deck.js and Extensions
Web Presentations, deck.js and Extensions
 
[Code night] natural language proccessing and machine learning
[Code night] natural language proccessing and machine learning[Code night] natural language proccessing and machine learning
[Code night] natural language proccessing and machine learning
 
孫民/從電腦視覺看人工智慧 : 下一件大事
孫民/從電腦視覺看人工智慧 : 下一件大事孫民/從電腦視覺看人工智慧 : 下一件大事
孫民/從電腦視覺看人工智慧 : 下一件大事
 

More from HITCON GIRLS

20200810_資訊洩漏所帶來的資安問題_turkey
20200810_資訊洩漏所帶來的資安問題_turkey20200810_資訊洩漏所帶來的資安問題_turkey
20200810_資訊洩漏所帶來的資安問題_turkey
HITCON GIRLS
 
CYBERSEC 2020 臺灣資安大會 windows malware hot 5 - 短短
CYBERSEC 2020 臺灣資安大會 windows malware hot 5  - 短短CYBERSEC 2020 臺灣資安大會 windows malware hot 5  - 短短
CYBERSEC 2020 臺灣資安大會 windows malware hot 5 - 短短
HITCON GIRLS
 
HITCON CTF 介紹 - HG 導覽活動
HITCON CTF 介紹 - HG 導覽活動HITCON CTF 介紹 - HG 導覽活動
HITCON CTF 介紹 - HG 導覽活動
HITCON GIRLS
 
HITCON CTF 2019 特色 - HG 導覽活動
HITCON CTF 2019 特色 - HG 導覽活動HITCON CTF 2019 特色 - HG 導覽活動
HITCON CTF 2019 特色 - HG 導覽活動
HITCON GIRLS
 
HITCON CTF 2017 特色 - HG 導覽活動
HITCON CTF 2017 特色 - HG 導覽活動HITCON CTF 2017 特色 - HG 導覽活動
HITCON CTF 2017 特色 - HG 導覽活動
HITCON GIRLS
 
HITCON CTF 2016 特色 - HG 導覽活動
HITCON CTF 2016 特色 - HG 導覽活動HITCON CTF 2016 特色 - HG 導覽活動
HITCON CTF 2016 特色 - HG 導覽活動
HITCON GIRLS
 
iThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdf
iThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdfiThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdf
iThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdf
HITCON GIRLS
 
Birds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom Kim
Birds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom KimBirds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom Kim
Birds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom Kim
HITCON GIRLS
 
Birds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - Howard
Birds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - HowardBirds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - Howard
Birds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - Howard
HITCON GIRLS
 
Birds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - Aaron
Birds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - AaronBirds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - Aaron
Birds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - Aaron
HITCON GIRLS
 
Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...
Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...
Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...
HITCON GIRLS
 
未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey
未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey
未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料
HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料
HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎
HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎
HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生
HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生
HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - Turkey
HITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - TurkeyHITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - Turkey
HITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - Turkey
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了
HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了
HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣
HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣
HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣
HITCON GIRLS
 
HITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - Belinda
HITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - BelindaHITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - Belinda
HITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - Belinda
HITCON GIRLS
 
逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)
逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)
逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)
HITCON GIRLS
 

More from HITCON GIRLS (20)

20200810_資訊洩漏所帶來的資安問題_turkey
20200810_資訊洩漏所帶來的資安問題_turkey20200810_資訊洩漏所帶來的資安問題_turkey
20200810_資訊洩漏所帶來的資安問題_turkey
 
CYBERSEC 2020 臺灣資安大會 windows malware hot 5 - 短短
CYBERSEC 2020 臺灣資安大會 windows malware hot 5  - 短短CYBERSEC 2020 臺灣資安大會 windows malware hot 5  - 短短
CYBERSEC 2020 臺灣資安大會 windows malware hot 5 - 短短
 
HITCON CTF 介紹 - HG 導覽活動
HITCON CTF 介紹 - HG 導覽活動HITCON CTF 介紹 - HG 導覽活動
HITCON CTF 介紹 - HG 導覽活動
 
HITCON CTF 2019 特色 - HG 導覽活動
HITCON CTF 2019 特色 - HG 導覽活動HITCON CTF 2019 特色 - HG 導覽活動
HITCON CTF 2019 特色 - HG 導覽活動
 
HITCON CTF 2017 特色 - HG 導覽活動
HITCON CTF 2017 特色 - HG 導覽活動HITCON CTF 2017 特色 - HG 導覽活動
HITCON CTF 2017 特色 - HG 導覽活動
 
HITCON CTF 2016 特色 - HG 導覽活動
HITCON CTF 2016 特色 - HG 導覽活動HITCON CTF 2016 特色 - HG 導覽活動
HITCON CTF 2016 特色 - HG 導覽活動
 
iThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdf
iThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdfiThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdf
iThome 資安大會 2019 駭客如何利用公開工具在內部網路中暢行無阻 - YCY.pdf
 
Birds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom Kim
Birds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom KimBirds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom Kim
Birds of a Feather 2017: 邀請分享 Light Up The Korean DarkWeb - Dasom Kim
 
Birds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - Howard
Birds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - HowardBirds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - Howard
Birds of a Feather 2017: 邀請分享 Glance into the Enterprise InfoSec Field - Howard
 
Birds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - Aaron
Birds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - AaronBirds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - Aaron
Birds of a Feather 2017: 邀請分享 IoT, SDR, and Car Security - Aaron
 
Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...
Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...
Birds of a Feather 2017: 邀請分享 Place of Attribution in Threat Intelligence - F...
 
未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey
未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey
未來大人物【互動工作坊】社群網路滲透生活,我們如何保護個資不外洩 蜘子珣&Turkey
 
HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料
HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料
HITCON GIRLS 資安萌芽推廣 2017: 從 CVE 找材料
 
HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎
HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎
HITCON GIRLS 資安萌芽推廣 2017: 你知道你連線的網站黑黑的嗎
 
HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生
HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生
HITCON GIRLS 資安萌芽推廣 2017: 密碼學深入淺出 阿毛 - 台科資管碩士生&圓圓 - 台科資管碩士生
 
HITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - Turkey
HITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - TurkeyHITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - Turkey
HITCON GIRLS 資安萌芽推廣 2017: 探索駭客的世界 - Turkey
 
HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了
HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了
HITCON GIRLS 資安萌芽推廣 2017: 還能不能好好玩遊戲了
 
HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣
HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣
HITCON GIRLS 資安萌芽推廣 2017: 從研究生轉職資安工程師 - 蜘子珣
 
HITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - Belinda
HITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - BelindaHITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - Belinda
HITCON GIRLS 資安萌芽推廣 2017: 資安工作很酷嘛?當然! - Belinda
 
逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)
逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)
逃避可恥還沒有用- 你不可不知的物聯網安全問題與挑戰(Ashley Shen & Belinda Lai)
 

Recently uploaded

一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
bakpo1
 
Gen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdfGen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdf
gdsczhcet
 
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
obonagu
 
Fundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptxFundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptx
manasideore6
 
Student information management system project report ii.pdf
Student information management system project report ii.pdfStudent information management system project report ii.pdf
Student information management system project report ii.pdf
Kamal Acharya
 
DfMAy 2024 - key insights and contributions
DfMAy 2024 - key insights and contributionsDfMAy 2024 - key insights and contributions
DfMAy 2024 - key insights and contributions
gestioneergodomus
 
Building Electrical System Design & Installation
Building Electrical System Design & InstallationBuilding Electrical System Design & Installation
Building Electrical System Design & Installation
symbo111
 
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
thanhdowork
 
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Dr.Costas Sachpazis
 
MCQ Soil mechanics questions (Soil shear strength).pdf
MCQ Soil mechanics questions (Soil shear strength).pdfMCQ Soil mechanics questions (Soil shear strength).pdf
MCQ Soil mechanics questions (Soil shear strength).pdf
Osamah Alsalih
 
14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application
SyedAbiiAzazi1
 
Railway Signalling Principles Edition 3.pdf
Railway Signalling Principles Edition 3.pdfRailway Signalling Principles Edition 3.pdf
Railway Signalling Principles Edition 3.pdf
TeeVichai
 
NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...
NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...
NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...
ssuser7dcef0
 
Final project report on grocery store management system..pdf
Final project report on grocery store management system..pdfFinal project report on grocery store management system..pdf
Final project report on grocery store management system..pdf
Kamal Acharya
 
PPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testingPPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testing
anoopmanoharan2
 
Immunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary AttacksImmunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary Attacks
gerogepatton
 
Understanding Inductive Bias in Machine Learning
Understanding Inductive Bias in Machine LearningUnderstanding Inductive Bias in Machine Learning
Understanding Inductive Bias in Machine Learning
SUTEJAS
 
6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)
ClaraZara1
 
Forklift Classes Overview by Intella Parts
Forklift Classes Overview by Intella PartsForklift Classes Overview by Intella Parts
Forklift Classes Overview by Intella Parts
Intella Parts
 
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
zwunae
 

Recently uploaded (20)

一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
一比一原版(SFU毕业证)西蒙菲莎大学毕业证成绩单如何办理
 
Gen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdfGen AI Study Jams _ For the GDSC Leads in India.pdf
Gen AI Study Jams _ For the GDSC Leads in India.pdf
 
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
 
Fundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptxFundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptx
 
Student information management system project report ii.pdf
Student information management system project report ii.pdfStudent information management system project report ii.pdf
Student information management system project report ii.pdf
 
DfMAy 2024 - key insights and contributions
DfMAy 2024 - key insights and contributionsDfMAy 2024 - key insights and contributions
DfMAy 2024 - key insights and contributions
 
Building Electrical System Design & Installation
Building Electrical System Design & InstallationBuilding Electrical System Design & Installation
Building Electrical System Design & Installation
 
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
 
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
 
MCQ Soil mechanics questions (Soil shear strength).pdf
MCQ Soil mechanics questions (Soil shear strength).pdfMCQ Soil mechanics questions (Soil shear strength).pdf
MCQ Soil mechanics questions (Soil shear strength).pdf
 
14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application
 
Railway Signalling Principles Edition 3.pdf
Railway Signalling Principles Edition 3.pdfRailway Signalling Principles Edition 3.pdf
Railway Signalling Principles Edition 3.pdf
 
NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...
NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...
NUMERICAL SIMULATIONS OF HEAT AND MASS TRANSFER IN CONDENSING HEAT EXCHANGERS...
 
Final project report on grocery store management system..pdf
Final project report on grocery store management system..pdfFinal project report on grocery store management system..pdf
Final project report on grocery store management system..pdf
 
PPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testingPPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testing
 
Immunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary AttacksImmunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary Attacks
 
Understanding Inductive Bias in Machine Learning
Understanding Inductive Bias in Machine LearningUnderstanding Inductive Bias in Machine Learning
Understanding Inductive Bias in Machine Learning
 
6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)
 
Forklift Classes Overview by Intella Parts
Forklift Classes Overview by Intella PartsForklift Classes Overview by Intella Parts
Forklift Classes Overview by Intella Parts
 
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
一比一原版(IIT毕业证)伊利诺伊理工大学毕业证成绩单专业办理
 

iThome 資安大會 2019 資安大會_AI 的逆襲,眼見真的為憑嗎? - Turkey&Melody.pdf

  • 2. 台灣駭客協會 秘書長 HITCON GIRLS 共同創辦⼈人 台灣駭客協會 專案經理理 HITCON GIRLS 成員 ABOUT US 2 Turkey Melody
  • 3. OUTLINE 3 • Deepfakes • Lyrebirds • 應⽤用 • 反制技術
  • 4. DEEPFAKES MAKES IT POSSIBLE TO CREATE FAKE VIDEOS OF ALMOST EVERYONE. 4
  • 5. 5 LET’S TALK ABOUT DEEPFAKES! 1. In December 2017, a user named “DeepFakes” announce the tool on Reddit Community of developers. 2. DeepFakes is a tool that utilizes deep learning to recognize and swap faces in pictures and videos. 5
  • 8. HOW DEEPFAKES WORK? >DeepFakes Tools 1. Fakeapp 2. DeepFaceLab 3. Fakeswap 4. Openfaceswap 5. Myfakeapp 8
  • 9. HOW DEEPFAKES WORK? >DeepFakes Tools 1. Fakeapp 2.DeepFaceLab 3. Fakeswap 4. Openfaceswap 5. Myfakeapp > 9
  • 11. HOW DEEPFAKES WORK? >Steps for: DeepFaceLab 2 31 Extract image from source video. Extract faces from source and target image. 4 Manually remove error image. 5 Model Training 6 Extract image from target video. Debug & Convert image to mp4 11
  • 12. HOW DEEPFAKES WORK? >Extract image from source&target video. FFmpeg FFmpeg is the leading multimedia framework, able to decode, encode, transcode, mux, demux, stream, filter and play pretty much anything that humans and machines have created. ffmpeg -i clipname -vf fps=framerate -qscale:v 2 "imagename%04d.jpg"Command_ 12
  • 13. HOW DEEPFAKES WORK? >Extract faces from source and target image (Ref: Facial landmarks with dlib, OpenCV, and Python) Divided a face into the following areas: 1. eyes(left/right) 2. eyebrows(left/right) 3. nose 4. chin 13
  • 14. HOW DEEPFAKES WORK? >Extract faces from source and target image (Ref: Real-Time Face Pose Estimation ) 14
  • 15. HOW DEEPFAKES WORK? >Extract faces from source and target image 15
  • 16. HOW DEEPFAKES WORK? >Extract faces from source and target image (Ref: One Millisecond Face Alignment with an Ensemble of Regression Trees)16
  • 17. HOW DEEPFAKES WORK? >Manually remove error image 17
  • 18. HOW DEEPFAKES WORK? >Manually remove error image 18
  • 19. HOW DEEPFAKES WORK? >Manually remove error image 19
  • 20. HOW DEEPFAKES WORK? >Model Training (Ref: https://deepfakes.com.cn/)20
  • 21. HOW DEEPFAKES WORK? >Model Training (Ref: https://deepfakes.com.cn/)21
  • 22. HOW DEEPFAKES WORK? >Model Training (Ref: https://deepfakes.com.cn/)22
  • 23. HOW DEEPFAKES WORK? >Model Training Models Types: 1. H64 2. H128 3. DF 4. LIAEF 5. LIAEF128YAW 6. MIAEF128 7. AVATAR 8. SAE H12823
  • 24. HOW DEEPFAKES WORK? >Model Training Models Types: 1. H64 2. H128 3. DF 4. LIAEF 5. LIAEF128YAW 6. MIAEF128 7. AVATAR 8. SAE H128 Train 越久效果越好! 24
  • 25. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 25
  • 26. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 26
  • 27. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 27
  • 28. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 28
  • 29. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 29
  • 30. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 30
  • 31. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 Debug Mode Use predicted mask? 1 Erosion (-100 to +100): (default = 0) 1 Seamless Erosion (0 to 40): (default = 0) 20 Blur (-200 to +200): (default = 0) 40 Hist-match threshold (0 to 255): (default = 255) default Face Scale (-50 to +50): (default = 0) default Transfer Color from predicted face? (LCT/RCT/no): (default = no) RCT Degrade Color Power of Final Image: (default = 0) default 31
  • 32. HOW DEEPFAKES WORK? >Debug & Convert image to mp4 ffmpeg -f image2 -i imagename%04d.jpg -vcodec libx264 -crf 15 -pix_fmt yuv420pCommand_ 32
  • 35. HOW DEEPFAKES WORK? >⼩小插曲 實測跑評:RTX 2080 ⼤大約需跑 6 個⼩小時以上會有不錯的成果 作者跪求 RTX 2080 Ti 中 XD 35
  • 36. HOW DEEPFAKES WORK? >不要讓貧窮限制你的 AI #讓 AI 聰明前 先飽滿你的⼝口袋 36
  • 39. CAN AI MIMIC YOUR VOICE? 39
  • 40. CAN YOU TELL A FAKE VOICE FROM A REAL ONE? 40 (Ref: https://lyrebird.ai/)
  • 41. • create a digitized version of your own voice. • Uses deep learning frameworks on AWS to develop and train its AI models What is Lyrebird? Record & Upload voice samples Train by DL Framework on AWS EC2 P3 Digital Voice Generated Optput audio of requested dialog (Ref: https://lyrebird.ai/)41
  • 42. LYREBIRD LIVE DEMO This is my honored to talk to everyone here. However, can you trust your eyes and ears to perceive reality? Machine learning can artificially mimic natural sounds to create digital voice. Just like me! 42
  • 43. Why my voice sounds awful? 43 (Ref: https://lyrebird.ai/)
  • 45. CHALLENGES FOR OUR FUTURE 45
  • 46. Why do you believe what I am telling you now? NEGATIVE EFFECTS 46
  • 47. POSITIVE EFFECTS Create images and voices from the past. 47 (Ref:apple) (Ref:Fast & Furious 6)
  • 49. FIGHT AGAINST FAKE VIDEOS! 49
  • 50. FAKE VIDEO DETECT SOLUTION >Detect malicious alterations with AI •abnormal compression signatures •lip sync analysis •Video metadata analysis •noise patterns analysis 50
  • 51. FAKE VIDEO DETECT SOLUTION > Fingerprint source videos & track provenance with blockchain smart contracts Create Hashing Blackchan record submitt 51
  • 52. REFERENCE • https://deepfakes.com.cn/ • https://www.youtube.com/watch?v=K98nTNjXkq8 • ttps://github.com/Fabsqrt/BitTigerLab • https://github.com/iperov/DeepFaceLab • https://www.youtube.com/watch?v=CR5Jr6Z1KI • https://lyrebird.ai/ • https://www.darpa.mil/ • https://medium.com/amber-video 52