Research map / Image generation / Image restoration
Neural Image Compression via Attentional Multi-Scale Back Projection and Frequency Decomposition
Ge Gao, Pei You, Rong Pan, Shunyuan Han, Yuanyuan Zhang, Yuchao Dai, Hojae Lee
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 AMBP up and down modules (ablation) | autoencoder codec · Kodak | BD-rate increase when removed: 8.61% | Improved |
| add soft selection in AMBP fusion over single channel attention weights (ablation) | AMBP · Kodak | BD-rate increase when removed: 1.21% | Improved |
| use AMBP down instead of 3-iteration DBPN (ablation) | AMBP · Kodak | BD-rate increase when removed: 1.17% | Improved |
| use AMBP up instead of 3-iteration DBPN (ablation) | AMBP · Kodak | BD-rate increase when removed: 1.98% | Improved |
| add frequency decomposition into base and detail layers (ablation) | encoder input · Kodak | BD-rate increase when removed: 2.08% | Improved |
| add dual attention module over four convolution layers (ablation) | latent fusion · Kodak | BD-rate increase when removed: 2.37% | Improved |
| add rounding loss to decoder finetuning (ablation) | finetuning stage · Kodak | BD-rate increase when removed: 3.52% | Improved |
| use the proposed model instead of reference model, costing more compute | Ours vs reference [10] · Kodak | parameters and latency: 25.4M vs 11.2M; encoding 104.24 vs 40.52 | Hurt |
Closest papers by meaning
- Linear Attention Modeling for Learned Image Compression (CVPR 2025 · similarity 0.88)
- Robustly overfitting latents for flexible neural image compression (NeurIPS 2024 · similarity 0.88)
- Diversify, Contextualize, and Adapt: Efficient Entropy Modeling for Neural Image Codec (NeurIPS 2024 · similarity 0.88)
- Controllable Distortion-Perception Tradeoff Through Latent Diffusion for Neural Image Compression (AAAI 2025 · similarity 0.87)
- GLIC: General Format Learned Image Compression (AAAI 2025 · similarity 0.87)
- Distribution-Aware Tensor Decomposition for Compression of Convolutional Neural Networks (NeurIPS 2025 · similarity 0.87)
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.
