Temporally coherent video anonymization through GAN inpainting
This work addresses privacy protection in video analysis by improving temporal consistency in anonymization, though it appears incremental as it builds on existing GAN and inpainting methods.
The paper tackles the problem of achieving temporally coherent face anonymization in video streams by proposing JaGAN, a two-stage system that masks faces and uses a Video GAN to inpaint them with generated faces, resulting in the introduction of a new video dataset and an Identity Invariance Score for quantifying coherency.
This work tackles the problem of temporally coherent face anonymization in natural video streams.We propose JaGAN, a two-stage system starting with detecting and masking out faces with black image patches in all individual frames of the video. The second stage leverages a privacy-preserving Video Generative Adversarial Network designed to inpaint the missing image patches with artificially generated faces. Our initial experiments reveal that image based generative models are not capable of inpainting patches showing temporal coherent appearance across neighboring video frames. To address this issue we introduce a newly curated video collection, which is made publicly available for the research community along with this paper. We also introduce the Identity Invariance Score IdI as a means to quantify temporal coherency between neighboring frames.