About this document
Paper 4 by dyica is a document available to read on EtoBox.
The paper introduces FaiResGAN, a novel Generative Adversarial Network designed for blind face restoration that preserves soft biometrics such as identity, ethnicity, age, and gender. It employs a pseudo-random batch composition algorithm to ensure fairness and mitigate bias, outperforming existing methods in both quantitative and qualitative assessments. User studies indicate a preference for FaiResGAN-restored images, highlighting its effectiveness in maintaining essential biometric attributes across dive
- Author
- dyica
- Language
- EN