Variable Rate Deep Image Compression With a Conditional Autoencoder

By Yoojin Choi, Mostafa El-Khamy, Jungwon Lee. 2019

280 citations recorded by the source corpus.

In this paper, we propose a novel variable-rate learned image compression framework with a conditional autoencoder. Previous learning-based image compression methods mostly require training separate networks for different compression rates so they can yield compressed images of varying quality. In contrast, we train and deploy only one variable-rate image compression network implemented with a conditional autoencoder. We provide two rate control parameters, i.e., the Lagrange multiplier and the quantization bin size, which are given as conditioning variables to the network. Coarse rate adaptation to a target is performed by changing the Lagrange multiplier, while the rate can be further fine-tuned by adjusting the bin size used in quantizing the encoded representation. Our experimental results show that the proposed scheme provides a better rate-distortion trade-off than the traditional variable-rate image compression codecs such as JPEG2000 and BPG. Our model also shows comparable and sometimes better performance than the state-of-the-art learned image compression models that deploy multiple networks trained for varying rates.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Differentiable Model Compression via Pseudo Quantization Noise (2021, methodology): “Additive noise was also studied in the context of image compression (Ballé, Laparra, and Simoncelli 2017; Choi, El-Khamy, and Lee 2019) in order to provide a differentiable pseudo-quantization operator.”
QARV: Quantization-Aware ResNet VAE for Lossy Image Compression (2023, methodology): “In an early work, Choi et al. introduce a conditional autoencoder with adjustable quantization bin size [46].”
Towards Backward-Compatible Continual Learning of Image Compression (2024, methodology): “To achieve variable-rate compression, we adopt the conditional convolution technique [12, 14], which applies an adaptive affine transformation to the convolu tional layer output (after layer normalization) to control the rate based on the input ,\.”
Scale-Space Flow for End-to-End Optimized Video Compression (2020, unknown): “Recent state-of-the-art models [40, 15, 26] use hyperprior-based architectures [8] with improvements including autoregressive context models [26] and multi-rate training [15].”
Checkerboard Context Model for Efficient Learned Image Compression (2021, methodology): “These context models are extended to more powerful and, of course, more computation-consuming ones by a series of later works [38, 19, 11, 10].”