Exploring AI-based Anonymization of Industrial Image and Video Data in the Context of Feature Preservation

TL;DR

This paper evaluates the effectiveness of DeepPrivacy2, a deep learning-based anonymization framework, for industrial images and videos, focusing on identity quality, temporal consistency, and usability for further analysis.

Contribution

It introduces the application of DeepPrivacy2 to industrial data and compares its performance with traditional anonymization methods.

Findings

DeepPrivacy2 produces high-quality, consistent anonymized identities.

The framework maintains temporal consistency in video data.

It preserves pose and action recognition capabilities after anonymization.

Abstract

With rising technologies, the protection of privacy-sensitive information is becoming increasingly important. In industry and production facilities, image or video recordings are beneficial for documentation, tracing production errors or coordinating workflows. Individuals in images or videos need to be anonymized. However, the anonymized data should be reusable for further applications. In this work, we apply the Deep Learning-based full-body anonymization framework DeepPrivacy2, which generates artificial identities, to industrial image and video data. We compare its performance with conventional anonymization techniques. Therefore, we consider the quality of identity generation, temporal consistency, and the applicability of pose estimation and action recognition.