Computer Graphics Seminar um olhar sobre os dados olmpicos - - PowerPoint PPT Presentation

computer graphics seminar
SMART_READER_LITE
LIVE PREVIEW

Computer Graphics Seminar um olhar sobre os dados olmpicos - - PowerPoint PPT Presentation

Computer Graphics Seminar um olhar sobre os dados olmpicos perceptions of olympic data Jlia Rabetti Giannella @juliagiannella juliagiannella@gmail.com OVERVIEW PART 1 OBERVATR!O2016 PART 2 Unfolding research: Deep Learning


slide-1
SLIDE 1

um olhar sobre os dados olímpicos perceptions of olympic data

Júlia Rabetti Giannella @juliagiannella juliagiannella@gmail.com

Computer Graphics Seminar

slide-2
SLIDE 2

OVERVIEW

  • PART 1

OBERVATÓR!O2016

  • PART 2

Unfolding research: Deep Learning experiences on Rio-2016

slide-3
SLIDE 3

PART 1 OBSERVATOR!O2016

slide-4
SLIDE 4

http://oo.impa.br

slide-5
SLIDE 5

X-RAY

  • March 2016 - December 2016
  • VISGRAF team
  • Products / publications

4 websites 1 blog + 1 twitter 4 presentations 1 technical report 1 digital publication 2 artistic exhibitions

slide-6
SLIDE 6

THE BEGINNING

interests

  • human activities from 


the perspective of 
 digital footprints

  • design as a tool for

comprehension of data through visualizations

visgraf

  • mathematical models /

creation of computational applications

  • New media / information

and communication technologies

slide-7
SLIDE 7

DIGITAL FOOTPRINTS

SOURCE

  • cellular network activity
  • credit card transactions
  • apps and websites usage
  • user generated content
  • sensor technology (IoT)

APPLICATIONS

  • marketing digital
  • urban planning
  • public policies (health, security)
  • data art
  • media studies
slide-8
SLIDE 8

SOURCE APPLICATIONS

Twitter media 
 (images and text) investigate the online debate about Rio 2016 how the Olympics are perceived and shared in social networks?

slide-9
SLIDE 9

IT FRAMEWORK

  • identify and represent the plurality of perceptions about Rio 2016
slide-10
SLIDE 10

DATA COLLECTION

  • Twitter API
  • via REST Queries
  • via Streaming
  • Scripts
  • Python
slide-11
SLIDE 11

MODELING

  • Categories
  • Rio-2016 aspects
  • Filters (Tweet)
  • User
  • User mention
  • Hashtag
  • Text
  • Time
slide-12
SLIDE 12

STORAGE

  • Database
  • Raw
  • OO
  • Stratification
  • Different levels
slide-13
SLIDE 13

SOFTWARE ARCHITECTURE

  • Server
  • Django / Mezzanine
  • Client
  • HTML 5 / CSS
  • Javascript
slide-14
SLIDE 14

VISUALIZATION

  • Specific aspects
  • Diferent frameworks
  • D3.js
  • Refining visual

aspects and interaction

slide-15
SLIDE 15

QUESTIONS

? ? ?

associations temporalities

slide-16
SLIDE 16

DEPLOYMENT

slide-17
SLIDE 17

DEPLOYMENT

beta 1.0 2.0 3.0

PAINEL DE TWEETS GALERIA DE IMAGENS ANÁLISE DE SENTIMENTO TOUR DA TOCHA OLHARES CONTROVERSOS MODALIDADES ESPORTIVAS RELAÇÕES PAÍS-ESPORTE MONITOR DE TEMAS MOSAICO DA TOCHA ATLETAS

slide-18
SLIDE 18

TWITTER

LISTAS MOMENTS BUSCA AVANÇADA

Source: https://twitter.com/

slide-19
SLIDE 19

TWEET PANEL

LINHA DO TEMPO (temporalidades) GRUPOS (associações) LINK HASHTAG USER MENTION

Source: http://oo.impa.br/dtweet/

slide-20
SLIDE 20

GROUPS OF ASSOCIATION

27 USUÁRIOS 100 HASHTAGS 66 HASHTAGS

slide-21
SLIDE 21

CONTROVERSIES (1st sketch)

slide-22
SLIDE 22

CONTROVERSIES (2nd sketch)

slide-23
SLIDE 23

CONTROVERSIES

Source: http://oo.impa.br/prenoms/#rio2016-1

slide-24
SLIDE 24

THEMES

slide-25
SLIDE 25

THEMES IDENTIFICATION

Source: http://oo.impa.br/temas_info/

slide-26
SLIDE 26

TERMS

slide-27
SLIDE 27

THEMES

Source: http://oo.impa.br/temas/

slide-28
SLIDE 28

PART 2 DEEP LEARNING AND RIO-2016

slide-29
SLIDE 29

180k images

slide-30
SLIDE 30

DEEP LEARNING

Source: https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/

slide-31
SLIDE 31

APPLICATIONS

  • Colorization of Black and White Images
  • Adding Sounds To Silent Movies
  • Image Classification and Object recognition *
  • Automatic Handwriting Generation
  • Automatic Video Generation *
  • Character Text Generation.
  • Image Caption Generation.
  • Automatic Game Playing
  • Artistic style transfer *
slide-32
SLIDE 32

IMAGE CLASSIFICATION

  • task: automatically classify and cluster images by subject

features related to the Olympic Games. Ex: Olympic Torch

  • CNN model
  • supervised learning (manually labeled 100 examples)
  • Inception-v3 CNN model *
  • TensorFlow (open source software library)**

* https://arxiv.org/abs/1409.4842 ** https://www.tensorflow.org/

slide-33
SLIDE 33

CONFIDENCE SCORE

Source: http://lvelho.impa.br/dl_rio2016/metodologia.html

A subset of 12 from 2091 images with confidence score over 83% for the Olympic torch category

slide-34
SLIDE 34

TORCH MOSAIC

Source: http://lvelho.impa.br/dl_rio2016/mosaico.html

slide-35
SLIDE 35

TORCH MOSAIC

Source: http://lvelho.impa.br/dl_rio2016/mosaico.html

slide-36
SLIDE 36

SPORTS CATEGORIES

  • Beach Volleyball
  • Cycling
  • Fencing
  • Soccer
  • Gymnastics
  • Judo
  • Golf
  • Basketball
  • Sailing
  • Tennis
  • Weightlifting
  • Equestrian
  • Field Hockey
  • Medals
  • Juan Martin Del Potro
  • Simone Biles
slide-37
SLIDE 37

CONFIDENCE SCORE

Source: http://lvelho.impa.br/dl_rio2016/metodologia.html

A subset of images with confidence score over 83% for the gymnastics category Training list

slide-38
SLIDE 38

AUTOMATIC SLIDESHOW GENERATION

  • App Photos in iOS 10
  • feature Moments

Source: http://lvelho.impa.br/dl_rio2016/metodologia.html

slide-39
SLIDE 39

VIDEO SPHERE

Source: http://lvelho.impa.br/dl_rio2016/videos.php

slide-40
SLIDE 40

PRODUCTS

website www.visgraf.impa.br/dl_rio2016 digital publishing folder

slide-41
SLIDE 41

INDISCIPLINAS

Presentation of video projections at Casa Franca-Brasil

poster

Source: http://lvelho.impa.br/dl_rio2016/indisciplinas.html Source: http://lvelho.impa.br/dl_rio2016/evento/apresentacao.mp4

video

slide-42
SLIDE 42

SUMMER PROGRAM AT IMPA

slide-43
SLIDE 43

... CAMPUS PARTY

CPBR10 Sao Paulo february, 4 11:45 - 12:45 Palco Criatividade

Source: http://campuse.ro/events/campus-party-brasil-2017/talk/observatoro2016-um-olhar-sobre-as-olimpiadas/

slide-44
SLIDE 44

OBRIGADA!

juliagiannella@gmail.com @juliagiannella