SDAKD: Student Discriminator Assisted Knowledge Distillation for Super-Resolution Generative Adversarial Networks

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kaparinos, Nikolaos, Mezaris, Vasileios
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911191953047552
author Kaparinos, Nikolaos
Mezaris, Vasileios
author_facet Kaparinos, Nikolaos
Mezaris, Vasileios
contents Generative Adversarial Networks (GANs) achieve excellent performance in generative tasks, such as image super-resolution, but their computational requirements make difficult their deployment on resource-constrained devices. While knowledge distillation is a promising research direction for GAN compression, effectively training a smaller student generator is challenging due to the capacity mismatch between the student generator and the teacher discriminator. In this work, we propose Student Discriminator Assisted Knowledge Distillation (SDAKD), a novel GAN distillation methodology that introduces a student discriminator to mitigate this capacity mismatch. SDAKD follows a three-stage training strategy, and integrates an adapted feature map distillation approach in its last two training stages. We evaluated SDAKD on two well-performing super-resolution GANs, GCFSR and Real-ESRGAN. Our experiments demonstrate consistent improvements over the baselines and SOTA GAN knowledge distillation methods. The SDAKD source code will be made openly available upon acceptance of the paper.
format Preprint
id arxiv_https___arxiv_org_abs_2510_03870
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SDAKD: Student Discriminator Assisted Knowledge Distillation for Super-Resolution Generative Adversarial Networks
Kaparinos, Nikolaos
Mezaris, Vasileios
Computer Vision and Pattern Recognition
Generative Adversarial Networks (GANs) achieve excellent performance in generative tasks, such as image super-resolution, but their computational requirements make difficult their deployment on resource-constrained devices. While knowledge distillation is a promising research direction for GAN compression, effectively training a smaller student generator is challenging due to the capacity mismatch between the student generator and the teacher discriminator. In this work, we propose Student Discriminator Assisted Knowledge Distillation (SDAKD), a novel GAN distillation methodology that introduces a student discriminator to mitigate this capacity mismatch. SDAKD follows a three-stage training strategy, and integrates an adapted feature map distillation approach in its last two training stages. We evaluated SDAKD on two well-performing super-resolution GANs, GCFSR and Real-ESRGAN. Our experiments demonstrate consistent improvements over the baselines and SOTA GAN knowledge distillation methods. The SDAKD source code will be made openly available upon acceptance of the paper.
title SDAKD: Student Discriminator Assisted Knowledge Distillation for Super-Resolution Generative Adversarial Networks
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.03870