Research map / Image generation / Image restoration
WeConvene: Learned Image Compression with Wavelet-Domain Convolution and Entropy Model
Haisheng Fu, Jie Liang, Zhenman Fang, Jingning Han, Feng Liang, Guohe Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add WeConv (Haar) to the baseline CNN codec (ablation) | baseline LIC · Kodak | BD-Rate vs VVC: -0.4% -> -4.7% | Improved |
| add WeChARM wavelet-domain entropy coding on top of WeConv (Haar) (ablation) | Baseline+WeConv · Kodak | BD-Rate vs VVC: -4.7% -> -8.2% | Improved |
| replace Haar with 5/3 wavelet (ablation) | WeConvene · Kodak | BD-Rate vs VVC: -8.2% -> -9.4% | Improved |
| replace 5/3 wavelet with 9/7 wavelet (ablation) | WeConvene · Kodak | BD-Rate vs VVC: -9.4% -> -9.8% | No clear effect |
| add WeConv (Haar) to baseline (ablation) | baseline LIC · Kodak | encoding time (s): 0.109s -> 0.110s | No clear effect |
| add two-step WeChARM entropy coding (ablation) | Baseline+WeConv · Kodak | decoding time (s): 0.147s -> 0.388s | Hurt |
| use 10 channel slices instead of 5 in WeChARM (ablation) | WeConvene (Haar) · Kodak low rate | PSNR (dB): 30.12 -> 30.16 | No clear effect |
| use 10 channel slices instead of 5 in WeChARM (ablation) | WeConvene (Haar) · Kodak | model size (MB): 107.15 -> 179.09 | Hurt |
| apply WeConvene (Haar) instead of best prior methods on PSNR | LIC · Kodak | PSNR gain over Zhu2022/Fu2023: about 0.2 dB | Improved |
| remove time-consuming SWAtten module in wavelet-domain ChARM (ablation) | WeChARM · Kodak | R-D performance | No clear effect |
Closest papers by meaning
- Learned Image Compression With Mixed Transformer-CNN Architectures (CVPR 2023 · similarity 0.89)
- Efficient Learned Image Compression without Entropy Coding (ICML 2026 · similarity 0.88)
- Few-Shot Domain Adaptation for Learned Image Compression (AAAI 2025 · similarity 0.87)
- Causal Context Adjustment Loss for Learned Image Compression (NeurIPS 2024 · similarity 0.87)
- On Disentangled Training for Nonlinear Transform in Learned Image Compression (ICLR 2025 · similarity 0.86)
- Linear Attention Modeling for Learned Image Compression (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
