Survey on Human Interactions

Alexandros Stergiou and I have compiled a comprehensive overview of literature on computer vision-based approaches to detect human-human interactions in videos, with a focus on recent CNN-based work. Under review, but already available on Arxiv!


Chapter in first Handbook on Social Signal Processing

My chapter on “Automatic Analysis of Bodily Social Signals” in the Handbook of Social Signal Processing is out! You can order the book through Amazon. Thanks to Judee Burgoon, Nadia Magnenat-Thalmann, Maja Pantic and Alessandro Vinciarelli for editing this comprehensive book!


Pain of Rejection Makes Us More Likely to Commit Fraud

In a user study into online insurance fraud, we found that people are more dishonest when they are rejected. It appears that the negative feeling ellicited by rejection, rather than a loss in monetary reward, drives this process. The research paper appears in Frontiers in Psychology, and already appears in the list of most downloaded papers.

The study has received internal press attention from Frontiers, ScienceDaily, NewsWise and others. We are looking for partners that deal with online fraud to test ways to prevent it.

Alejandro Moreno successfully defended his PhD thesis

On April 21, Alejandro Moreno successfully defended his PhD thesis “From Traditional to Interactive Playspaces: Automatic Analysis of Player Behavior in the Interactive Tag Playground”. Congratulations!2016_ale_defense

He will now continue this exciting work as a postdoctoral researcher at the University of Twente. In the mean time, check out the accepted Entertainment Computing paper on subjective and objective evaluation of technology-enhanced playgrounds.

Interactive Playground in JMUI


An open-access paper by Alejandro Moreno and me is published at the Journal on Multimodal User Interfaces (JMUI). Download it here. We discuss how to automate the analysis of behavior of playing children. We compare traditional tag with an interactive version played in our Interactive Playground. We also introduce two algorithms for the detection of tagger and runner roles by looking at interactions between children.