Objective
The objective of the project is to create a database of stickers located in the district of Gotico, Barcelona by collecting geolocated photos, analyzing & sorting them by machine learning and further research to then create a framework for evaluating and improving local economy, artists and political movements marketing strategies.
Strategy
By collecting geolocated photos, one photo per sticker, a database is created for further processing by a machine learning algorithm that is supposed to analyse the frequency of a sticker aswell as it’s type – plain text, plain graphic, graphic + text and QR codes. In further manual research the background – either political, commercial and other – and the creators and their agenda are evaluated.
Analysis
Gotico is part of the “Ciutat Vella” district used to be a roman village with narrow, winding streets that create quite a labyrinth and high housing density.The famous “las ramblas”, the town hall of Barcelona and the Plaça Reial are located in Gotico and provide high economic and touristic density aswell as bars, restaurants gathered around. Prior to collecting the photos of the stickers the street network aswell as it’s infrastructure were analysed.
Bias
When taking the pictures the “Bias” of the upcoming dataset became obvious – how to make sure we covered all streets and plazas? With the tracking the walks through google maps! Some of the stickers also were out of reach to make proper photos and also photos put on posts with small diameters were “unreadable” and distorted and therefore not part of the dataset. The biggest bias remains the factor of time – as the collecting was distributed within a period of three weeks, the dataset is fragmented as most of the stickers get scratched of within a short period of time.
Collecting the stickers
Training set
A sample of 100 photos were taken to form a “training” set for the development of the analysis algorithm. Type and background of the training photos were evaluated and added manually.
Density & Frequency
Type & Background
The analysis of type and background uncovered that the majority of the stickers were actually “text & graphic” – “plain text”, “plain graphic” and “qr code” are distributed quite even. Also its’s not a surprise that for most of the stickers there was no information to find out it’s background or creator. Also no surprise is that the majority of the stickers background is political as there is a high proximity to several municipal institutions in Gotico.
Creators & Agenda
For the three most frequent stickers the creators and their agenda were identified. A graphic designer and graffiti artist who enganges in climate change activism, a local company owner to promote his transport business aswell as little motivational messages created in Germany. Obviously 2 of the 3 creators are not located in Spain wich reflects the high diversity of of different nationalities visiting Gotico and Barcelona.
Potentials?
Are there potential uses or benefits the created dataset could offer? There could be a platform, with features like “vote the best sticker” – connecting sticker artists around the world or there could be a way to evaluate the best locations to place stickers according to their purpose – maybe this was just a very fun project to shed some light on the very diverse sticker culture in Barcelona.
