Analysis of Human Action Recognition Features in Person Identification Systems for Anti-Bullying Applications

Perelli, Gianpaolo;Micheletto, Marco;Orru', Giulia;Avvisati, Giulia;Capozza, Massimo;Luca Marcialis, Gian

2024-01-01

---

Abstract

This study examines the privacy implications of using Human Activity Recognition (HAR) features for violent action detection in the fight against bullying. Given the sensitive nature of such applications, these systems must comply with privacy regulations, such as the General Data Protection Regulation (GDPR). Our analysis focuses on whether the features commonly employed in HAR systems comply with privacy regulations, particularly regarding the risk of inadvertently exposing personal identities when the data includes sensitive biometric information. The findings, derived from the development of a personal recognition system based on movement data captured by mobile device sensors, underscore the need for stringent privacy safeguards in HAR systems.