際際滷

際際滷Share a Scribd company logo
1/15
丶个 丼 亊 丼豫乂丱
舒仍亠亳亶 亞亳仆
Data Science Meetup
7 亠于舒仍 2018 亞仂亟舒
2/15
弌亠亞仂亟仆 仂弍亟亳仄
1 ARVI Lab
2 仆仂于仆舒 亳亟亠
3 仂仂亠仆亳亠 仄仂亟亠仍亳
4 亳仂于从舒
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
仂 500+ 磲亠 CPU
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
仂 500+ 磲亠 CPU
仂仆仂 600 TFLOPS
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
仂 500+ 磲亠 CPU
仂仆仂 600 TFLOPS
仆仂亞仂 3+ TB RAM
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
仂 500+ 磲亠 CPU
仂仆仂 600 TFLOPS
仆仂亞仂 3+ TB RAM
亠仆 仄仆仂亞仂 200+ TB 亟亳从仂于仂亞仂
仗仂舒仆于舒
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
仂 500+ 磲亠 CPU
仂仆仂 600 TFLOPS
仆仂亞仂 3+ TB RAM
亠仆 仄仆仂亞仂 200+ TB 亟亳从仂于仂亞仂
仗仂舒仆于舒
仂仂 亰舒于舒仍亳 200 揃 103
磲亠 GPU
3/15
1 ARVI Lab
亠亟仂舒于仍磳仄 亠 亟仍 AI 仗仂亠从仂于 亳 于仂亰仄仂亢仆仂
舒亢亳仂于从亳 于 从仂仄仗舒仆亳亳
亳亳亠 舒亠 弌亠ミ
k.stetsiuk@arvilab.com
仂 500+ 磲亠 CPU
仂仆仂 600 TFLOPS
仆仂亞仂 3+ TB RAM
亠仆 仄仆仂亞仂 200+ TB 亟亳从仂于仂亞仂
仗仂舒仆于舒
仂仂 亰舒于舒仍亳 200 揃 103
磲亠 GPU
4/15
2 仆仂于仆舒 亳亟亠
丱仂亳仄 于亳亟亠 仗舒仆仂于 仗仂 弍亳亰仆亠 于 于亳舒仍仆仂仄 仄亳亠
5/15
2 仆仂于仆舒 亳亟亠
丱仂亳仄 亢亳 于 于亳舒仍仆仂仄 仄亳亠
6/15
2 仆仂于仆舒 亳亟亠
仂亢亠, 仆亳仄舒  仗仂仄仂 仗亠亳舒仍仆 从舒仄亠?
Fusion4D: Real-time Performance Capture of Challenging Scenes (2016)
Mingsong Dou, Sameh Khamis, Yury Degtyarev, Philip Davidson, Sean Ryan Fanello, Adarsh
Kowdle, Sergio Orts Escolano, Christoph Rhemann, David Kim, Jonathan Taylor, Pushmeet
Kohli, Vladimir Tankovich, Shahram Izadi
7/15
2 仆仂于仆舒 亳亟亠
亢仆仂 仗亠亠仆仂亳 仄亳仄亳从 仆舒 亰舒舒仆亠亠 仗仂亟亞仂仂于仍亠仆仆 仄仂亟亠仍
8/15
3 仂仂亠仆亳亠 仄仂亟亠仍亳
弌从舒仆亳亠仄 亞仂仍仂于 亠仍仂于亠从舒
8/15
3 仂仂亠仆亳亠 仄仂亟亠仍亳
亠仄 舒仍仂仆仆 仄仂亟亠仍
8/15
3 仂仂亠仆亳亠 仄仂亟亠仍亳
亠仄 仂仂于亠于亳亠 仄亠亢亟 于亠亳仆舒仄亳 亟于 仄仂亟亠仍亠亶
9/15
3 仂仂亠仆亳亠 仄仂亟亠仍亳
亠亰仍舒
9/15
3 仂仂亠仆亳亠 仄仂亟亠仍亳
亠亰仍舒 仂亠仍仂仄仍磳
10/15
3 仂仂亠仆亳亠 仄仂亟亠仍亳
舒亠仄 于 仂?
11/15
4 亳仂于从舒
仂亞仂于舒 仂仗亳仄亳亰舒亳仂仆仆舒 亰舒亟舒舒
2
揃 s
2
+
iI
q 揃 R 揃 Mi 揃 s + b  ki
2
 min
q,k,R,b,s
q  仄舒舒弍
R  仄舒亳舒 仗仂于仂仂舒
Mi  仆舒弍仂 从仂仂亟亳仆舒 仂从亳 i 于 舒亰仆 仄仂亟亠仍 (仄舒亳舒)
b  于亠从仂 仄亠亠仆亳
s  仆舒弍仂 于亠仂于 亟仍 从舒亢亟仂亶 仄仂亟亠仍亳 (于亠从仂)
ki  仂从舒 从舒仆亳仂于舒仆仆仂亶 仄仂亟亠仍亳, 从仂仂舒 弍仍亳亢亠 于亠 亟亞亳 从 q 揃 R 揃 Mi 揃 s + b
ki = arg min
tT
t  q 揃 R 揃 Mi 揃 s + b
2
12/15
4 亳仂于从舒
亠亞仍亳亰舒亳
仍亳 仗仂仗亳 2
揃 s
2
, 仂 于 仂-舒仄亠 亰舒亞亳 仂亟亳仆 于舒亰亳仄仂亟仂
13/15
4 亳仂于从舒
丐舒亟舒仄!
14/15
弌仗舒亳弍仂 亰舒 于仆亳仄舒仆亳亠!
舒从 亰仆舒 弍仂仍亠?
仂亳舒 舒亳
A Morphable Model For The Synthesis Of 3D Faces (1999)
Volker Blanz and Thomas Vetter
FaceWarehouse: a 3D Facial Expression Database for Visual Computing (2012)
Chen Cao, Yanlin Weng, Shun Zhou, Yiying Tong, Kun Zhou
Markov Chain Monte Carlo for Automated Face Image Analysis (2017)
Sandro Sch即onborn, Bernhard Egger, Andreas Morel-Forster, Thomas Vetter
Learning a model of facial shape and expression from 4D scans (2017)
Tianey Li, Timo Bolkart, Michael J. Black, Hao Li, Javier Romero
14/15
弌仗舒亳弍仂 亰舒 于仆亳仄舒仆亳亠!
舒从 亰仆舒 弍仂仍亠?
仂亳舒 舒亳
A Morphable Model For The Synthesis Of 3D Faces (1999)
Volker Blanz and Thomas Vetter
FaceWarehouse: a 3D Facial Expression Database for Visual Computing (2012)
Chen Cao, Yanlin Weng, Shun Zhou, Yiying Tong, Kun Zhou
Markov Chain Monte Carlo for Automated Face Image Analysis (2017)
Sandro Sch即onborn, Bernhard Egger, Andreas Morel-Forster, Thomas Vetter
Learning a model of facial shape and expression from 4D scans (2017)
Tianey Li, Timo Bolkart, Michael J. Black, Hao Li, Javier Romero
 
Fusion4D: Real-time Performance Capture of Challenging Scenes (2016)
Mingsong Dou, Sameh Khamis, Yury Degtyarev, Philip Davidson, Sean Ryan Fanello, Adarsh
Kowdle, Sergio Orts Escolano, Christoph Rhemann, David Kim, Jonathan Taylor, Pushmeet
Kohli, Vladimir Tankovich, Shahram Izadi
15/15
弌仗舒亳弍仂 亰舒 于仆亳仄舒仆亳亠!
舒从 亰仆舒 弍仂仍亠?
弌仗仂亳  仄亠仆
valeriy.krygin at gmail.com
facebook.com valerij.krygin

More Related Content

Realistic VR experience requires visualization of realistic players, which includes the transfer of body movements and emotions in real time