EPFL people search via profile photo comparisons using machine learning techniques


In this project, we are interested in developing a brand new search framework for navigating through the database of EPFL people via comparing their profile photos.

Project idea

  • Problem: let’s say you remember someone you’ve seen on campus but you don’t know his/her name. It is not very time-efficient to just check every single profile on people.epfl.ch one by one until you find that person.

  • Solution: instead we ask you to compare two faces and say which of the two looks most like the one you have mind. Then we give you another pair of candidates and so on until we show you the face of the person you were looking for.


As the first step we will be working with the dataset of current public profile pictures of EPFL people. Later on, we might expand this database to all EPFL people.


  1. Collect the dataset of public photos of EPFL people
  2. Build up a web page for comparison-based search
  3. Integrate current machine learning techniques into the search engine


• Web programming • Solid knowledge of linear algebra and probability • Python and machine learning skills is a plus