On the Adaptability of Person Re-Identification Models

Person Re-Identification is a computer vision task that aims at retrieving a target person across multiple, non overlapping cameras. It is a central component of any intelligent surveillance system, and has seen a steady increase in research efforts. Despite the problem's considerable difficulty, state-of-the-art models have achieved impressive results over all benchmark datasets, which has motivated us to investigate the applicability of such models in a real-world scenario. To this end, we present a brief overview of the most important aspects of person re-identification, as well as some experimental results regarding the domain gap problem of person re-identification models, which to this day remains the main challenge to the adoption of such models in a real-world setting.