What does your Instagram say about you? Exploring Google Cloud Vision

by Christine

Slide 1

Slide 1 text

@tech_christine @ryanhos What does your Christine Seeman and Ryan Hochstetler say about you?

Slide 2

Slide 2 text

@tech_christine @ryanhos What is in an image?

Slide 3

Slide 3 text

@tech_christine @ryanhos What is in an image?

Slide 4

Slide 4 text

@tech_christine @ryanhos What do you focus on?

Slide 5

Slide 5 text

@tech_christine @ryanhos What do you focus on?

Slide 6

Slide 6 text

@tech_christine @ryanhos What is in it that you see?

Slide 7

Slide 7 text

@tech_christine @ryanhos What is in it that you see?

Slide 8

Slide 8 text

@tech_christine @ryanhos But what does an application see in them?

Slide 9

Slide 9 text

@tech_christine @ryanhos Meet Google Cloud Vision

Slide 10

Slide 10 text

@tech_christine @ryanhos

Slide 11

Slide 11 text

@tech_christine @ryanhos

Slide 12

Slide 12 text

@tech_christine @ryanhos Little bit different vision

Slide 13

Slide 13 text

@tech_christine @ryanhos Cloud Vision API Landmarks Face detection Image labeling Optical character recognition Explicit content detection Logo detection Object localization Crop hint detection

Slide 14

Slide 14 text

@tech_christine @ryanhos But how can we use it?

Slide 15

Slide 15 text

@tech_christine @ryanhos Let's look at the client library

Slide 16

Slide 16 text

@tech_christine @ryanhos Let's combine accounts with

Slide 17

Slide 17 text

@tech_christine @ryanhos

Slide 18

Slide 18 text

@tech_christine @ryanhos Onto the code

Slide 19

Slide 19 text

@tech_christine @ryanhos image_annotator_client.rb

Slide 20

Slide 20 text

@tech_christine @ryanhos image_annotator_client.rb

Slide 21

Slide 21 text

@tech_christine @ryanhos image_annotator_client.rb

Slide 22

Slide 22 text

@tech_christine @ryanhos image_annotator_client.rb

Slide 23

Slide 23 text

@tech_christine @ryanhos So that got us from this

Slide 24

Slide 24 text

@tech_christine @ryanhos To this

Slide 25

Slide 25 text

@tech_christine @ryanhos analysis_reader.rb

Slide 26

Slide 26 text

@tech_christine @ryanhos analysis_reader.rb

Slide 27

Slide 27 text

@tech_christine @ryanhos After histogram generation Labels Landmarks Logos

Slide 28

Slide 28 text

@tech_christine @ryanhos Then finally the tag cloud

Slide 29

Slide 29 text

@tech_christine @ryanhos Let's explore the different detections

Slide 30

Slide 30 text

@tech_christine @ryanhos Face detection

Slide 31

Slide 31 text

@tech_christine @ryanhos What’s a “gnathion”?

Slide 32

Slide 32 text

@tech_christine @ryanhos raster_annotator.rb

Slide 33

Slide 33 text

@tech_christine @ryanhos raster_annotator.rb

Slide 34

Slide 34 text

@tech_christine @ryanhos Sentiment Detection Probably should be called “Sentiment Best Guestimation”

Slide 35

Slide 35 text

@tech_christine @ryanhos Joy

Slide 36

Slide 36 text

@tech_christine @ryanhos Surprise

Slide 37

Slide 37 text

@tech_christine @ryanhos Anger?

Slide 38

Slide 38 text

@tech_christine @ryanhos Sorrow?

Slide 39

Slide 39 text

@tech_christine @ryanhos Image Composition

Slide 40

Slide 40 text

@tech_christine @ryanhos Label detection

Slide 41

Slide 41 text

@tech_christine @ryanhos @expertvagabond

Slide 42

Slide 42 text

@tech_christine @ryanhos @expertvagabond

Slide 43

Slide 43 text

@tech_christine @ryanhos @ripleyandrue

Slide 44

Slide 44 text

@tech_christine @ryanhos @ripleyandrue

Slide 45

Slide 45 text

@tech_christine @ryanhos Safe Search detection

Slide 46

Slide 46 text

@tech_christine @ryanhos Yes it knows if it is a hotdog

Slide 47

Slide 47 text

@tech_christine @ryanhos ...and when it's not

Slide 48

Slide 48 text

@tech_christine @ryanhos Seems accurate…

Slide 49

Slide 49 text

@tech_christine @ryanhos Oh, there might be a bit of bias…

Slide 50

Slide 50 text

@tech_christine @ryanhos Violence is difficult too…

Slide 51

Slide 51 text

@tech_christine @ryanhos Racy but not violent?!?

Slide 52

Slide 52 text

@tech_christine @ryanhos Doctored images are hit and miss We chose not to include the image of a former US president photoshopped to look like a dictator and war criminal. Google labeled it “Possibly” a spoof; i.e. altered to be funny or oﬀensive

Slide 53

Slide 53 text

@tech_christine @ryanhos Text detection

Slide 54

Slide 54 text

@tech_christine @ryanhos @letterfolk

Slide 55

Slide 55 text

@tech_christine @ryanhos @letterfolk "MY\nBRAIN\nHAS\nTO0\nMANY\nTABS\nOPEN\nDechtlatte\nTHIS WEEK\npt de C\nCAit s pg\n" MY BRAIN HAS TO0 MANY TABS OPEN Dechtlatte THIS WEEK pt de C CAit s pg

Slide 56

Slide 56 text

@tech_christine @ryanhos @letterfolk

Slide 57

Slide 57 text

@tech_christine @ryanhos @letterfolk YOURE NOT THE\nBOSS OF ME..\nIWHISPER UNDER\nMY BREATH AS I\nCLEAN UP ALL OF\nMY CHILDREN'S\nMESSES.\nTRE\n YOURE NOT THE BOSS OF ME.. IWHISPER UNDER MY BREATH AS I CLEAN UP ALL OF MY CHILDREN'S MESSES. TRE

Slide 58

Slide 58 text

@tech_christine @ryanhos @letterfolk

Slide 59

Slide 59 text

@tech_christine @ryanhos @letterfolk Seth\nElizabeth\nMat\nIT'S BEGINNING TO\namazon Prime\nLOOK A LOT LIKE...\nvmazon Prime\necho\nI GOT MY MONEY'S\nWORTH FROM MY\n1t\nare en\na ton Prime\nAMAZON PRIME\nMEMBERSHIP\nNany\necho\neciro\n Seth Elizabeth Mat IT'S BEGINNING TO amazon Prime LOOK A LOT LIKE... vmazon Prime echo I GOT MY MONEY'S WORTH FROM MY 1t\nare en a ton Prime AMAZON PRIME MEMBERSHIP Nany echo eciro

Slide 60

Slide 60 text

@tech_christine @ryanhos Logo detection

Slide 61

Slide 61 text

@tech_christine @ryanhos

Slide 62

Slide 62 text

@tech_christine @ryanhos @nike

Slide 63

Slide 63 text

@tech_christine @ryanhos @nike

Slide 64

Slide 64 text

@tech_christine @ryanhos @nike 1. nike 2. nike plus 3. nike azul 4. blue nike 5. nike store

Slide 65

Slide 65 text

@tech_christine @ryanhos @christine_seeman

Slide 66

Slide 66 text

@tech_christine @ryanhos @christine_seeman

Slide 67

Slide 67 text

@tech_christine @ryanhos @christine_seeman 1. etelä suomen sanomat 2. rogue status 3. american horror story 4. delta skymiles 5. deník !"!"

Slide 68

Slide 68 text

@tech_christine @ryanhos Landmark detection

Slide 69

Slide 69 text

@tech_christine @ryanhos @visit_nebraska

Slide 70

Slide 70 text

@tech_christine @ryanhos @iloveny

Slide 71

Slide 71 text

@tech_christine @ryanhos @iloveny

Slide 72

Slide 72 text

@tech_christine @ryanhos @expertvagabond

Slide 73

Slide 73 text

@tech_christine @ryanhos Have you been to London, NE?

Slide 74

Slide 74 text

@tech_christine @ryanhos The British Museum there is lovely

Slide 75

Slide 75 text

@tech_christine @ryanhos British Museum, alternate photo

Slide 76

Slide 76 text

@tech_christine @ryanhos So much alike

Slide 77

Slide 77 text

@tech_christine @ryanhos This wasn’t even a building…

Slide 78

Slide 78 text

@tech_christine @ryanhos But what does my instagram say about me?

Slide 79

Slide 79 text

@tech_christine @ryanhos

Slide 80

Slide 80 text

@tech_christine @ryanhos @christineseeman

Slide 81

Slide 81 text

@tech_christine @ryanhos Blooper cloud

Slide 82

Slide 82 text

@tech_christine @ryanhos Where to go from here? What to do next with Google Cloud Vision

Slide 83

Slide 83 text

@tech_christine @ryanhos @visit_nebraska

Slide 84

Slide 84 text

@tech_christine @ryanhos Vs @visitcalifornia

Slide 85

Slide 85 text

@tech_christine @ryanhos Vs @iloveny

Slide 86

Slide 86 text

@tech_christine @ryanhos Can we teach a machine to know if a pic is from NY vs CA vs NE

Slide 87

Slide 87 text

@tech_christine @ryanhos Explore bias in Machine Learning?

Slide 88

Slide 88 text

@tech_christine @ryanhos How much will this cost you? For us, about $65

Slide 89

Slide 89 text

@tech_christine @ryanhos Technology Used •Ruby (No Rails) •Rspec •Ruby Vision API Client Libraries •Google Cloud Storage •Rmagick (Ruby binding to Imagemagick) •MagicCloud tag cloud gem

Slide 90

Slide 90 text

@tech_christine @ryanhos Reference Links •https://cloud.google.com/vision/ •https://cloud.google.com/vision/docs/ •https://googleapis.github.io/google-cloud-ruby/docs/ •https://github.com/GoogleCloudPlatform/ruby-docs- samples/blob/master/vision/quickstart.rb •https://github.com/zverok/magic_cloud

Slide 91

Slide 91 text

@tech_christine @ryanhos All the instagram accounts* @ripleyandrue @visit_nebraska @wolffolins @expertvagabond @iloveny @letterfolk @myraswim @visitcalifornia @nationalportraitgallery @nike * and none were harmed in the making of this talk