Pose-Based Sign Language Appearance Transfer
This addresses privacy concerns in sign language processing by obfuscating identity, though it is incremental as it builds on existing pose-based rendering techniques.
The paper tackles the problem of transferring a signer's appearance in sign language skeletal poses while preserving sign content, using estimated poses to maintain natural movements and transitions. The method reduces signer identification accuracy but slightly harms sign recognition performance, highlighting a tradeoff between privacy and utility.
We introduce a method for transferring the signer's appearance in sign language skeletal poses while preserving the sign content. Using estimated poses, we transfer the appearance of one signer to another, maintaining natural movements and transitions. This approach improves pose-based rendering and sign stitching while obfuscating identity. Our experiments show that while the method reduces signer identification accuracy, it slightly harms sign recognition performance, highlighting a tradeoff between privacy and utility. Our code is available at https://github.com/sign-language-processing/pose-anonymization.