• Stato del progetto

    Terminato

  • Tipo di progetto

    Progetto competitivo

  • Data di inizio

    13 ottobre 2022

  • Data di fine

    13 aprile 2023

Fully homomorphic encryption (FHE) systems allow the performing of arbitrary computations on ciphered data without ever decrypting them. As such, they represent a game changer technology in many different scenarios, including that of cloud computing. In fact, FHE allows cloud providers to process the data of customers while still guaranteeing their privacy. In this project, Fully Homomorphic Encryption For Computer Vision (FHE4CV), we focus on the problem of detecting the presence of humans in encrypted images. Our main goal is to develop a prototype implementation of a privacy-preserving human detection system, which will take in input an image and output a binary value (e.g., one if a person is detected, and zero otherwise). The main drawback of FHE is its high run-time complexity, which still hinders its widespread application. In this project, we will explore the extent to which current FHE technologies are already suitable for realizing a human detection system with an acceptable run-time. In case this will not be possible, trade-offs between privacy and computational complexity will be studied, and a feasible solution will be proposed accordingly.

Responsabile del progetto
  • Davide Andreoletti