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

Luca Rossi, Eric Boscaro, Andrea Torsello

Research output: Chapter in Book/Published conference outputConference publication


The last decade has seen a huge expansion in the use of social media to extract data about human behaviour. While metadata and textual information have taken the lion's share as data sources for social media analysis, geotagged image-based platforms represent an unprecedented and as yet almost untapped source of data to analyse human behaviour and characterise the physical space we live in. In this paper we investigate the use of Instagram photos to analyse tourism consumption. We take the city of Venice (Italy) as a case study and we collect a dataset of about 90k photos taken between January 2014 and December 2015. Using computer vision techniques, we build a supervised classifier which assigns each photo to one of six different categories. We then observe how the frequency and spatial distribution of these categories varies with time. This in turn allows us to confirm the existence of a number of touristic hotspots associated with different events, such as Venice Carnival and Biennale. Our analysis also uncovers the existence of touristic flows associated with these events, such as the Folklore Line that marks the path of tourists from "Santa Lucia" railway station to "San Marco" square during the Carnival period. Overall, our findings confirm the effectiveness of the proposed framework to investigate tourism consumption using Instagram data.
Original languageEnglish
Title of host publication WWW '18 Companion of the The Web Conference 2018 on The Web Conference 2018
ISBN (Print) 978-1-4503-5640-4
Publication statusPublished - 23 Apr 2018
EventCompanion of the The Web Conference 2018 - Lyon, France
Duration: 23 Apr 201827 Apr 2018


ConferenceCompanion of the The Web Conference 2018


Dive into the research topics of 'Venice through the Lens of Instagram: A Visual Narrative of Tourism in Venice'. Together they form a unique fingerprint.

Cite this