Don’t call it privacy-preserving or human-centric pose estimation if you don’t measure privacy

Michele Baldassini, Francesco Pistolesi, Beatrice Lazzerini

Keywords: Artificial Intelligence, Computer Vision, Ethics, Human-centric, Human Pose Estimation, Privacy

TL;DR: This paper argues that privacy must be measured and integrated as a core dimension in the evaluation of human pose estimation systems.

Abstract: This position paper argues that human pose estimation (HPE) cannot be considered privacy-preserving or human-centric unless privacy is measured and evaluated. Although privacy concerns have become more visible in recent years, HPE systems are still assessed almost exclusively using accuracy metrics. Privacy is neither defined in measurable terms nor linked to regulatory requirements, and common deployment architectures introduce additional risks due to data transmission and storage. We highlight the limitations of current practices, including the continued reliance on RGB inputs and the lack of benchmarks that reflect legal and ethical constraints. We call for a shift in evaluation practices: privacy must become part of how HPE systems are designed, tested, and compared.

Lay Summary:

Human pose estimation (HPE) systems track how people move by identifying the position of their body parts in images or videos. These technologies have become highly accurate and are now used in fields such as healthcare, sports analysis, and surveillance. However, HPE systems often rely on detailed visual data, transmit them to servers, and store them in ways that can expose personal information, creating privacy risks.

Currently, there are no shared metrics or benchmarks to assess how well these systems protect individuals’ identities, nor standards that reflect ethical or legal expectations.

In this paper, we call for a shift: privacy should be treated as a core performance dimension. By making privacy measurable—alongside accuracy—researchers and developers can create systems that are both effective and respectful of the people they observe.

Submission Number: 736