Accelerating Learnt Video Codecs with Gradient Decay and Layer-wise Distillation

TL;DR

This paper introduces a model-agnostic pruning method using gradient decay and layer-wise distillation to significantly reduce computational complexity in learned video codecs while maintaining compression performance.

Contribution

It proposes a novel pruning scheme that enhances efficiency of learned video codecs with minimal accuracy loss, applicable across multiple models.

Findings

Up to 65% reduction in MACs

2x speed-up in decoding

Less than 0.3dB drop in BD-PSNR

Abstract

In recent years, end-to-end learnt video codecs have demonstrated their potential to compete with conventional coding algorithms in term of compression efficiency. However, most learning-based video compression models are associated with high computational complexity and latency, in particular at the decoder side, which limits their deployment in practical applications. In this paper, we present a novel model-agnostic pruning scheme based on gradient decay and adaptive layer-wise distillation. Gradient decay enhances parameter exploration during sparsification whilst preventing runaway sparsity and is superior to the standard Straight-Through Estimation. The adaptive layer-wise distillation regulates the sparse training in various stages based on the distortion of intermediate features. This stage-wise design efficiently updates parameters with minimal computational overhead. The proposed approach has been applied to three popular end-to-end learnt video codecs, FVC, DCVC, and DCVC-HEM. Results confirm that our method yields up to 65% reduction in MACs and 2x speed-up with less than 0.3dB drop in BD-PSNR. Supporting code and supplementary material can be downloaded from: https://jasminepp.github.io/lightweightdvc/