A Practical Approach to Face Verification: An End-to-End Implementation

The aim of this thesis is to present the project that was carried out during the internship at DuckMa Srl. The objective was to develop a Proof of Concept for a face verification system that would encompass three core functionalities: data collection and preprocessing, the ability to determine whether a person present in an image or video feed matches the vector selected from a data structure, and the removal of a specific person, given their facial representation. Furthermore, it was essential to ascertain the feasibility of operating a face verification system on devices with constrained computational resources.