Internships at ReSnap

ReSnap is the 4th fastest growing tech company of The Netherlands, based in the city center of Nijmegen. We developed an AI that can predict someone’s “best” photos and automatically generate a photo book. Best is not simply an aesthetics score, but the real emotional value of a photo. Our consumer solution is the easiest photo book solution worldwide.

Determining and using user features from photo sets at ReSnap

For a personalised selection of photos, user features can be very important. However, this kind of information is not easy to retrieve. We want to explore the possibility to determine user features from photo sets and apply these user features to our selection algorithms. User features are for example gender or age range, but more user features can be thought of that might affect a personalised selection. The input data are sets of photos users upload to create photo books with. We have a labelled set available for age range and gender. Other features and unsupervised methods to segment users should also be considered.

The project consists of multiple goals:

  • Determine user features that are important for a personalised selection using the dataset and our photo selection pipeline
  • Research the viability to cluster users in different segment based on photo sets and the usefulness of these clusters in our photo selection pipeline
  • Use a ML method (probably a deep learning network) to extract the above features from photo sets
  • This algorithm needs to run in realtime. The goal is to optimise for accuracy, speed and size
  • Report increase of the selection score by adding different user features to the selection pipeline

Come join us for this internship and work together to make a worldwide impact!

Contact: Thomas Beguin (ReSnap) or Arjen P. de Vries (RU)