Research map / Image generation / Image restoration
Cassic: Towards Content-Adaptive State-Space Models for Learned Image Compression
Shiyu Qin, Jinpeng Wang, Yimin Zhou, Bin Chen, Tianci Luo, Baoyi An, Tao Dai, Shu-Tao Xia, Yaowei Wang
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 CASS on the encoder side instead of 2DSS (ablation) | Cassic nonlinear transform · Kodak | BD-rate: 2.67% | Improved |
| add CASS on the decoder side instead of 2DSS (ablation) | Cassic nonlinear transform · Kodak | BD-rate: 4.91% | Improved |
| add CASS on both sides instead of 2DSS (ablation) | Cassic nonlinear transform · Kodak | BD-rate: 6.15% | Improved |
| add CASS on both sides instead of 2DSS (ablation) | Cassic nonlinear transform · CLIC Professional Validation | BD-rate: 7.21% | Improved |
| unfold high-value positions first in CASS rather than inverse order (ablation) | Cassic CASS · Kodak | BD-rate: 1.34% | Improved |
| add CASS in entropy model instead of four-directional scanning (ablation) | Cassic entropy model · Kodak | BD-rate: 1.79% | Improved |
| use CA-VSSB entropy model instead of Mixed (SWAtten) block (ablation) | Cassic entropy model · Kodak | BD-rate: 2.39% | Improved |
| use CA-VSSB entropy model instead of CCA (NAF) block (ablation) | Cassic entropy model · Kodak | BD-rate: 4.27% | Improved |
| use Cassic instead of MambaVC | Mamba-based image compression · Kodak | BD-rate / latency: 1.49% BD-rate gain; latency 282 -> 261 ms | Improved |
| use Cassic instead of CCA (CNN) | image compression · Kodak | BD-rate / latency: 2.17% BD-rate gain; latency 201 -> 261 ms | Mixed |
| use Cassic instead of MLIC++ | image compression · CLIC Professional Validation | BD-rate: 0.9% | Mixed |
Closest papers by meaning
- MambaIC: State Space Models for High-Performance Learned Image Compression (CVPR 2025 · similarity 0.92)
- CALLIC: Content Adaptive Learning for Lossless Image Compression (AAAI 2025 · similarity 0.87)
- Diversify, Contextualize, and Adapt: Efficient Entropy Modeling for Neural Image Codec (NeurIPS 2024 · similarity 0.86)
- Laplacian-guided Entropy Model in Neural Codec with Blur-dissipated Synthesis (CVPR 2024 · similarity 0.86)
- Linear Attention Modeling for Learned Image Compression (CVPR 2025 · similarity 0.85)
- Causal Context Adjustment Loss for Learned Image Compression (NeurIPS 2024 · similarity 0.85)
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.
