跳到主要导航 跳到搜索 跳到主要内容

ON DISENTANGLED TRAINING FOR NONLINEAR TRANSFORM IN LEARNED IMAGE COMPRESSION

  • Han Li
  • , Shaohui Li*
  • , Wenrui Dai*
  • , Maida Cao
  • , Nuowen Kan
  • , Chenglin Li
  • , Junni Zou
  • , Hongkai Xiong
  • *此作品的通讯作者
  • Shanghai Jiao Tong University
  • Tsinghua University

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

摘要

Learned image compression (LIC) has demonstrated superior rate-distortion (RD) performance compared to traditional codecs, but is challenged by training inefficiency that could incur more than two weeks to train a state-of-the-art model from scratch. Existing LIC methods overlook the slow convergence caused by compacting energy in learning nonlinear transforms. In this paper, we first reveal that such energy compaction consists of two components, i.e., feature decorrelation and uneven energy modulation. On such basis, we propose a linear auxiliary transform (AuxT) to disentangle energy compaction in training nonlinear transforms. The proposed AuxT obtains coarse approximation to achieve efficient energy compaction such that distribution fitting with the nonlinear transforms can be simplified to fine details. We then develop wavelet-based linear shortcuts (WLSs) for AuxT that leverages wavelet-based downsampling and orthogonal linear projection for feature decorrelation and subband-aware scaling for uneven energy modulation. AuxT is lightweight and plug-and-play to be integrated into diverse LIC models to address the slow convergence issue. Experimental results demonstrate that the proposed approach can accelerate training of LIC models by 2 times and simultaneously achieves an average 1% BD-rate reduction. To our best knowledge, this is one of the first successful attempt that can significantly improve the convergence of LIC with comparable or superior rate-distortion performance. Code will be released at https://github.com/qingshi9974/AuxT.

源语言英语
主期刊名13th International Conference on Learning Representations, ICLR 2025
出版商International Conference on Learning Representations, ICLR
页84718-84738
页数21
ISBN(电子版)9798331320850
出版状态已出版 - 2025
已对外发布是
活动13th International Conference on Learning Representations, ICLR 2025 - Singapore, 新加坡
期限: 24 4月 2025 → 28 4月 2025

丛书

姓名13th International Conference on Learning Representations, ICLR 2025

会议

会议13th International Conference on Learning Representations, ICLR 2025
国家/地区新加坡
市Singapore
时期24/04/25 → 28/04/25

学术指纹

探究 'ON DISENTANGLED TRAINING FOR NONLINEAR TRANSFORM IN LEARNED IMAGE COMPRESSION' 的科研主题。它们共同构成独一无二的学术指纹。

引用此