Contextual Privacy Preservation in Active and Assisted Living Using Omnidirectional Cameras

This paper explores contextual privacy preservation using top-view omnidirectional dioptric cameras, particularly within Active and Assisted Living (AAL) environments. These technologies hold significant promise for enhancing the quality of life for older adults and individuals with disabilities, yet they also introduce substantial privacy challenges. Using the concept of "Privacy by Context" based on Nissenbaum’s theory of contextual integrity, this paper proposes a privacy-preserving pipeline for omnidirectional cameras that aligns with EU legal frameworks. As part of the research, the ODIN dataset was created to facilitate the creation of machine learning models aimed at scene understanding with omnidirectional dioptric camera images.