Distilling Complexity-Scalable Learned Image Compression Models via Neural Architecture Search
IEEE Transactions on Circuits and Systems for Video Technology · 2026-07 · 36 (7) · 9660-9675
DOI: 10.1109/tcsvt.2026.3670313 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Circuits and Systems for Video Technology, 2026-07. Verfasst von Shen Wang und 6 weiteren Autor:innen. Die Arbeit nennt 77 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Fundamental Research Funds for the Central Universities, Shanghai Key Laboratory of Digital Media Processing and Transmission.
Abstract (Originalsprache)
Learned image compression (LIC) has attracted considerable attention due to its outstanding rate-distortion performance at the cost of high computational complexity. However, most existing LIC methods with symmetric, fixed-complexity encoder-decoder frameworks fail to accommodate varying encoding/decoding complexity constraints across application scenarios, which poses significant deployment challenges. To address this, we propose the Complexity-Scalable Learned Image Compression (CSLIC) method. It leverages neural architecture search (NAS) and knowledge distillation to construct multiple encoder and decoder configurations with varying complexity levels, achieving complexity scalability and superior complexity-compression trade-offs. Specifically, we first construct a search space encompassing building block types and channel dimensions and introduce module-wise NAS algorithm into LIC task. Through parameter-shared supernet training and sampling, we identify the optimal synthesis transform and analysis transform architecture under different complexity constraints. Finally, combined with the knowledge distillation guided progressive training strategy, we implement a complexity-scalable learned image compression framework. Experimental results demonstrate that our proposed method achieves superior compression performance-complexity trade-offs, delivering BD-rate savings of -13.59% to -22.88% with computational complexities ranging from 97.65 to 283.54 KMACs/pixel. At the same complexity level, our method significantly outperforms existing SOTA methods while offering flexible adjustment of encoding and decoding complexity.
Förderung
- National Natural Science Foundation of China
- Fundamental Research Funds for the Central Universities
- Shanghai Key Laboratory of Digital Media Processing and Transmission
- Higher Education Discipline Innovation Project
- Ant Group Research Fund
Zitieren
APA
Wang, S., Cheng, Z., Feng, D., Wang, Q., Gu, Q., Song, L., & Zhang, W. (2026). Distilling Complexity-Scalable Learned Image Compression Models via Neural Architecture Search. IEEE Transactions on Circuits and Systems for Video Technology, 36(7), 9660-9675. https://doi.org/10.1109/tcsvt.2026.3670313
Vancouver
Wang S, Cheng Z, Feng D, Wang Q, Gu Q, Song L, et al. Distilling Complexity-Scalable Learned Image Compression Models via Neural Architecture Search. IEEE Transactions on Circuits and Systems for Video Technology. 2026;36(7):9660-9675. doi:10.1109/tcsvt.2026.3670313
BibTeX
@article{Wang2026,
author = {Wang, Shen and Cheng, Zhengxue and Feng, Donghui and Wang, Qi and Gu, Qunshan and Song, Li and Zhang, Wenjun},
title = {Distilling Complexity-Scalable Learned Image Compression Models via Neural Architecture Search},
journal = {IEEE Transactions on Circuits and Systems for Video Technology},
year = {2026},
volume = {36},
number = {7},
pages = {9660--9675},
doi = {10.1109/tcsvt.2026.3670313}
}
EUVIMED