Research map / Image generation / Fast diffusion sampling
Towards Efficient Low-rate Image Compression with Frequency-aware Diffusion Prior Refinement
Yichong Xia, Yimin Zhou, Jinpeng Wang, Bin Chen
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 Consistency Refinement Estimator (CRE) (ablation) | DiffCR · CLIC20 | BD-rate %: up to 40% | Improved |
| replace cross-attention with Frequency Decoupling Attention (FDA) (ablation) | DiffCR · CLIC20 | BD-rate %: approximately 14% | Improved |
| add second stage training (ablation) | DiffCR · CLIC20 | BD-rate %: around 47% | Improved |
| use mixed semantic embeddings instead of pure text embeddings (ablation) | DiffCR · CLIC20 | BD-rate %: 9% | Improved |
| add FDA (with CRE) (ablation) | DiffCR · CLIC20 | BD-rate %: 13.52 -> 0 (LPIPS), 15.64 -> 0 (FID) | Improved |
| add semantic embedding (ablation) | DiffCR · CLIC20 | BD-rate %: 8.32 -> 0 (LPIPS), 10.11 -> 0 (FID) | Improved |
| add 2-stage training (ablation) | DiffCR · CLIC20 | BD-rate %: 44.5 -> 0 (LPIPS), 51.1 -> 0 (FID) | Improved |
| two-step decoding versus 50-step diffusion baselines | DiffCR · Kodak | decoding speed: over 10x faster | Improved |
Closest papers by meaning
- OSCAR: One-Step Diffusion Codec Across Multiple Bit-rates (NeurIPS 2025 · similarity 0.92)
- Ultra Lowrate Image Compression with Semantic Residual Coding and Compression-aware Diffusion (ICML 2025 · similarity 0.90)
- Decouple Distortion from Perception: Region Adaptive Diffusion for Extreme-low Bitrate Perception Image Compression (CVPR 2025 · similarity 0.90)
- Lossy Compression with Pretrained Diffusion Models (ICLR 2025 · similarity 0.90)
- UltraDiff:Transferring High-Fidelity Priors to Compressed Latent Spaces for High-resolution Image Generation (NeurIPS 2026 · similarity 0.89)
- One-Step Diffusion-Based Image Compression with Semantic Distillation (NeurIPS 2025 · similarity 0.89)
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.
