Jianing Lu is a Chinese undergraduate student studying physics at the University of Toronto. She loves reading sci-fi, drawing, watching plays and volunteering. She is engaged in many multidisciplinary artistic processes, including creating visual artwork and theatre producing. She is trying to combine all these interests in her projects.
Project: Instagram Artwork Classification through a Neural Network
To classify contemporary artwork from several Instagram art accounts and analyze how the images’ features relate to their popularity (number of likes and number of comments).
Main Results in Detail
The images downloaded from several Instagram accounts were classified according to their features extracted by a neural network. There is an obvious, visually identifiable difference between clusters but some deviations still remain. Features of the most popular images appear to be very random, which is saying the relation between popularity and image features is relatively random. Other metadata information aside from features might influence popularity greatly.
It is worth trying to make further popularity predictions on Instagram posts and to analyze if other metadata information (length of captions, etc.) has an important influence on popularity.