Image-based profiling
Images/photos of users is one of the highly-available
personal on-line information. People often publish images on
different social media (e.g., Facebook or Instagram) without disclosing their
private attributes such as their interests, list of friends, or places that
they visited. Although it is usually fairly easy for humans to infer private
information (e.g., age, sex, interests, locations, etc.) about the image owner by carefully analyzing different objects of the image
(faces, background objects, activities, etc.) [1], the automation of such privacy attacks had been
difficult before the advent of Deep Learning.
The task is to process images of different users and infer their private
information (e.g., age, sex, interests, locations, etc.) using machine learning.
Required skills: programming (preferably python)
Preferred (but not required) skills: machine learning (neural networks, deep learning)
[1] http://index.hu/fortepan/2017/01/08/itt_vannak_a_bravuros_fortepan_megfejtesek/