Wavelet Knowledge Distillation: Towards Efficient Image-to-Image Translation
This addresses the problem of deploying efficient GANs for practitioners in image-to-image translation, though it is incremental as it builds on existing knowledge distillation methods.
The paper tackles the inefficiency and high memory usage of state-of-the-art GANs in image-to-image translation by proposing wavelet knowledge distillation, which focuses on high-frequency bands to improve small GANs, achieving 7.08 times compression and 6.80 times acceleration on CycleGAN with almost no performance drop.
Remarkable achievements have been attained with Generative Adversarial Networks (GANs) in image-to-image translation. However, due to a tremendous amount of parameters, state-of-the-art GANs usually suffer from low efficiency and bulky memory usage. To tackle this challenge, firstly, this paper investigates GANs performance from a frequency perspective. The results show that GANs, especially small GANs lack the ability to generate high-quality high frequency information. To address this problem, we propose a novel knowledge distillation method referred to as wavelet knowledge distillation. Instead of directly distilling the generated images of teachers, wavelet knowledge distillation first decomposes the images into different frequency bands with discrete wavelet transformation and then only distills the high frequency bands. As a result, the student GAN can pay more attention to its learning on high frequency bands. Experiments demonstrate that our method leads to 7.08 times compression and 6.80 times acceleration on CycleGAN with almost no performance drop. Additionally, we have studied the relation between discriminators and generators which shows that the compression of discriminators can promote the performance of compressed generators.