Instagram for museums

Classification of images posted on Instagram by museum visitors (UK, US, KR, FR).

– 2020 –

with Boa Rhee and Gangta Choi (Chung-Ang University, South Korea)

We used Google Vision machine learning models to identify objects represented in Instagram posts about museum exhibitions. We then trained a Word2Vec model on the labels assigned to the images and clustered them in seven recurring topics. Beside making an original research contribution, we shared the output of our image classification and a simplified version of the workflow so that museum staff could use it to autonomously monitor their audience online presence.

Tools and skills

Python, Google Vision API, Word2Vec, Flourish

Web scraping, image classification, machine learning, data visualisation

Outputs

Rhee, Boa, Federico Pianzola, and Gangta Choi. 2021 (in press). “Analyzing the museum experience through the lens of Instagram posts”. Curator: The Museum Journal.

Labels’ dataset: https://zenodo.org/record/3706462#.YDEspC9Q36A