Self-Organized Variational Autoencoders (Self-VAE) for Learned Image Compression
Abstract
In end-to-end optimized learned image compression, it is standard practice to use a convolutional variational autoencoder with generalized divisive normalization (GDN) to transform images into a latent space. Recently, Operational Neural Networks (ONNs) that learn the best non-linearity from a set of alternatives, and their self-organized variants, Self-ONNs, that approximate any non-linearity via Taylor series have been proposed to address the limitations of convolutional layers and a fixed nonlinear activation. In this paper, we propose to replace the convolutional and GDN layers in the variational autoencoder with self-organized operational layers, and propose a novel self-organized variational autoencoder (Self-VAE) architecture that benefits from stronger non-linearity. The experimental results demonstrate that the proposed Self-VAE yields improvements in both rate-distortion performance and perceptual image quality.
Cite
@article{arxiv.2105.12107,
title = {Self-Organized Variational Autoencoders (Self-VAE) for Learned Image Compression},
author = {M. Akın Yılmaz and Onur Keleş and Hilal Güven and A. Murat Tekalp and Junaid Malik and Serkan Kıranyaz},
journal= {arXiv preprint arXiv:2105.12107},
year = {2021}
}
Comments
Accepted for publication in IEEE International Conference on Image Processing (ICIP) 2021