Flexible Mixed Precision Quantization for Learned Image Compression

TL;DR

This paper introduces a flexible mixed precision quantization method for learned image compression models, optimizing layer-wise bit-widths to reduce computational costs while maintaining high coding performance.

Contribution

It proposes a novel adaptive search algorithm for layer-wise bit-width assignment, improving resource utilization over fixed-precision quantization in LIC models.

Findings

Achieves better BD-Rate performance under fixed model size constraints.

Reduces computational complexity of LIC models through optimized quantization.

Provides open-source implementation for reproducibility.

Abstract

Despite its improvements in coding performance compared to traditional codecs, Learned Image Compression (LIC) suffers from large computational costs for storage and deployment. Model quantization offers an effective solution to reduce the computational complexity of LIC models. However, most existing works perform fixed-precision quantization which suffers from sub-optimal utilization of resources due to the varying sensitivity to quantization of different layers of a neural network. In this paper, we propose a Flexible Mixed Precision Quantization (FMPQ) method that assigns different bit-widths to different layers of the quantized network using the fractional change in rate-distortion loss as the bit-assignment criterion. We also introduce an adaptive search algorithm which reduces the time-complexity of searching for the desired distribution of quantization bit-widths given a fixed model size. Evaluation of our method shows improved BD-Rate performance under similar model size constraints compared to other works on quantization of LIC models. We have made the source code available at gitlab.com/viper-purdue/fmpq.