Venice through the Lens of Instagram: A Visual Narrative of Tourism - - PowerPoint PPT Presentation

▶
venice through the lens of instagram a visual narrative
SMART_READER_LITE
LIVE PREVIEW

Venice through the Lens of Instagram: A Visual Narrative of Tourism - - PowerPoint PPT Presentation

Oct 28, 2023 95 likes •608 views

Venice through the Lens of Instagram: A Visual Narrative of Tourism in Venice Luca Rossi 1 , Eric Boscaro 2 , Andrea Torsello 2 1. Aston University, United Kingdom 2. Universit Ca Foscari Venezia, Italy The 8th International Workshop on

slide-1
SLIDE 1

Venice through the Lens of Instagram:
 A Visual Narrative of Tourism in Venice

Luca Rossi1, Eric Boscaro2, Andrea Torsello2

  • 1. Aston University, United Kingdom
  • 2. Università Ca’ Foscari Venezia, Italy

The 8th International Workshop on Location and the Web, 24 April 2018, Lyon, France

slide-2
SLIDE 2
  • The tourism industry has boasted virtually uninterrupted growth over time
  • International tourist arrivals have increased from 25 million globally in 1950

to 278 million in 1980, 674 million in 2000, and 1,235 million in 2016

Source: https://www.e-unwto.org/doi/pdf/10.18111/9789284419029

slide-3
SLIDE 3
  • Tourists make an increasing use of photo-sharing social media like

Instagram and Flickr to share their experiences online

  • Geotagged data provides a rich source of information to study tourism

consumption

slide-4
SLIDE 4
  • The city of Venice (Italy) provides an interesting case study, being one of

the most popular destinations in one of the world most visited countries

Source: http://blog.euromonitor.com/2016/01/top-100-city-destinations-ranking-2016.html

slide-5
SLIDE 5

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories
slide-6
SLIDE 6

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Lagoon

slide-7
SLIDE 7

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Townscape

slide-8
SLIDE 8

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Art

slide-9
SLIDE 9

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Folklore

slide-10
SLIDE 10

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Food

slide-11
SLIDE 11

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Other

slide-12
SLIDE 12

Dataset

  • We retrieve 90,000 geotagged Instagram photos

taken in Venice from Jan 2014 to Dec 2015

  • We group these images into 6 categories

Lagoon Townscape Art Folklore Food Other

slide-13
SLIDE 13

Image classification framework

  • We create a training set of 600 manually annotated

images, 100 per class

  • With this training set, we classify the remaining

images using a combination of SIFT features, BOW representations and SVM classifiers

http://www.ics.uci.edu/~majumder/VC/211HW3/vlfeat/doc/overview/sift.html

slide-14
SLIDE 14

Image classification framework

  • We create a training set of 600 manually annotated

images, 100 per class

  • With this training set, we classify the remaining

images using a combination of SIFT features, BOW representations and SVM classifiers

http://www.ics.uci.edu/~majumder/VC/211HW3/vlfeat/doc/overview/sift.html

slide-15
SLIDE 15

Image classification framework

  • We create a training set of 600 manually annotated

images, 100 per class

  • With this training set, we classify the remaining

images using a combination of SIFT features, BOW representations and SVM classifiers

http://vgg.fiit.stuba.sk/2015-02/bag-of-visual-words-in-opencv/

slide-16
SLIDE 16

Image classification framework

  • We create a training set of 600 manually annotated

images, 100 per class

  • With this training set, we classify the remaining

images using a combination of SIFT features, BOW representations and SVM classifiers

  • We perform 5-fold cross validation to compute the

average classification accuracy of the classifier (68%)

slide-17
SLIDE 17

1st classifier: confusion matrix

Lagoon 7ownVFape Art Folklore Food VarLeV Lagoon 7ownVFape Art Folklore Food VarLeV 73 11 2 6 8 11 72 2 8 7 1 12 71 4 1 11 10 11 64 7 8 2 5 2 86 5 3 15 12 16 12 42

  • Low misclassification

rate for every class except Other – Over 50% misclassification rate for this class!

slide-18
SLIDE 18

1st classifier: confusion matrix

Lagoon 7ownVFape Art Folklore Food VarLeV Lagoon 7ownVFape Art Folklore Food VarLeV 73 11 2 6 8 11 72 2 8 7 1 12 71 4 1 11 10 11 64 7 8 2 5 2 86 5 3 15 12 16 12 42

  • We can reduce this

issue by artificially increasing the probability of assigning an image to Other – Multiply prob of assigning to
 Other by M

slide-19
SLIDE 19

Lagoon 7ownVFape Art Folklore Food 9arLeV Lagoon 7ownVFape Art Folklore Food 9arLeV 68 5 1 4 2 20 9 68 1 7 15 7 65 28 8 8 55 5 24 4 77 19 9 4 2 6 79

2nd classifier: confusion matrix

  • M is optimised

through 5-fold cross- validation on the training set

  • Optimal value is

found to be M=2

slide-20
SLIDE 20

Misclassified photos: example

Other (wrong) Townscape (wrong)

slide-21
SLIDE 21

Misclassified photos: example

Townscape (correct) Folklore (correct)

slide-22
SLIDE 22

Categories distribution

Lagoon TownVFape Art FolNlore Food VarLeV 5000 10000 15000 20000 25000 30000 1umber oI ImageV 2014 2015

slide-23
SLIDE 23

Categories distribution

Lagoon TownVFape Art FolNlore Food VarLeV 5000 10000 15000 20000 25000 30000 1umber oI ImageV 2014 2015

In both years, about 50% of the images are assigned to Other, for a total of 44k out of 90k images

slide-24
SLIDE 24

Categories distribution

Lagoon TownVFape Art FolNlore Food VarLeV 5000 10000 15000 20000 25000 30000 1umber oI ImageV 2014 2015

25% of the photos are in the Townscape category, which comprises architectural elements such as bridges,
 churches, squares, highlighting the rich architectural heritage of Venice

slide-25
SLIDE 25

Number of photos taken over time

The trend is in line with the worldwide growth in Instagram active users,
 which have more than doubled from the beginning of 2014 to the end of 2015

J D n F e b D r A S r D y J u n J u O A u J S e S 2 F t 1

  • v

D e F 1000 1500 2000 2500 3000 3500 1umber oI ImDJes 2014 2015

slide-26
SLIDE 26

Number of photos taken over time

Peak corresponding to Carnival 2015

J D n F e b D r A S r D y J u n J u O A u J S e S 2 F t 1

  • v

D e F 1000 1500 2000 2500 3000 3500 1umber oI ImDJes 2014 2015

slide-27
SLIDE 27

Frequency of Art photos in 2015

The shaded area shows the increase in the frequency of Art photos during the 56th Art Biennale

JDn Feb 0Dr ASr 0Dy Jun JuO AuJ 6eS 2Ft 1ov DeF 10 12 14 16 18 20 % oI Art ImDJes

slide-28
SLIDE 28

Frequency of Art photos in 2015

Note also the slight increase during the Carnival period.
 This may be due to the increased number of cultural events


  • rganised in museums and galleries during that period

JDn Feb 0Dr ASr 0Dy Jun JuO AuJ 6eS 2Ft 1ov DeF 10 12 14 16 18 20 % oI Art ImDJes

slide-29
SLIDE 29

Heatmap: Lagoon 2014

slide-30
SLIDE 30

Heatmap: Lagoon 2015

slide-31
SLIDE 31

Heatmap: Townscape 2014

slide-32
SLIDE 32

Heatmap: Townscape 2015

slide-33
SLIDE 33

Heatmap: Art 2014

slide-34
SLIDE 34

Heatmap: Art 2014

Guggenheim Biennale Biennale

slide-35
SLIDE 35

Heatmap: Art 2015

slide-36
SLIDE 36

Heatmap: Art 2015

Guggenheim Biennale Biennale San Giorgio Maggiore Palazzo Grassi

slide-37
SLIDE 37

Heatmap: Folklore 2014

slide-38
SLIDE 38

Heatmap: Folklore 2015

slide-39
SLIDE 39

Heatmap: Folklore 2015

Rialto bridge San Marco square Santa Lucia (railway station)

slide-40
SLIDE 40

Heatmap: Food 2014

slide-41
SLIDE 41

Heatmap: Food 2014

Al Paradiso Perduto (restaurant) Al Timon Bragozzo (restaurant) Tonolo (pastry shop)

slide-42
SLIDE 42

Heatmap: Food 2015

slide-43
SLIDE 43

Heatmap: February 2015 (during Carnival)

slide-44
SLIDE 44

Heatmap: March 2015 (after Carnival)

slide-45
SLIDE 45

SDntD CroFe SDn 0DrFo CDstello Dorsoduro SDn Polo CDnnDregio 5 10 15 20 25 % oI Folklore IPDges During CDrnivDl AIter CDrnivDl

During Carnival After Carnival

slide-46
SLIDE 46

SDntD CroFe SDn 0DrFo CDstello Dorsoduro SDn Polo CDnnDregio 5 10 15 20 25 % oI Folklore IPDges During CDrnivDl AIter CDrnivDl

During Carnival After Carnival

slide-47
SLIDE 47
  • 1. Study limited to users of Instagram
  • 2. Image classification not state-of-the-art
  • 3. Large number of photos classified as Other
  • 4. Textual information (e.g, #hashtags) discarded
slide-48
SLIDE 48

Conclusion and future work

  • We explored tourism consumption through the lens
  • f Instagram
  • The analysis of 90k photos over two years highlights

the presence of touristic hotspots

  • The signal is influenced by external events and can

reveal preferred touristic routes during such events

slide-49
SLIDE 49

Conclusion and future work

  • Potential areas of applications:

– Urban planning – Marketing and advertising campaigns – Personalised tourist guide by linking city representation to user preferences, as determined by his/her shared photos

  • Future work will investigate text to associate

sentiment to places and will use CNN to improve image classification

slide-50
SLIDE 50

Questions?

https://cs.aston.ac.uk/~rossil/ l.rossi@aston.ac.uk blextar

The 8th International Workshop on Location and the Web, 24 April 2018, Lyon, France