Research map / Image generation / Image restoration
Implicit Neural Representations for Image Compression
Yannick Strümpler, Janis Postels, Ren Yang, Luc Van Gool, Federico Tombari
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 meta-learned initialization to INR compression (ablation) | basic INR compression (SIREN + positional encoding) · Kodak convergence | epochs to reach equal performance: 25000 -> 2500 epochs (90% less training time) | Improved |
| add meta-learned initialization to INR compression (ablation) | basic INR compression · Kodak convergence (early training) | PSNR vs epochs: 3 epochs (meta-learned) beats 50 epochs (basic) | Improved |
| add meta-learned initialization (rate-distortion) (ablation) | basic INR compression · Kodak and CelebA | rate-distortion (PSNR vs bpp) | Improved |
| add meta-learned initialization (rate-distortion) (ablation) | basic INR compression · Kodak 2x and 4x reduced resolution | rate-distortion (PSNR vs bpp) | Improved |
| use sine activation instead of ReLU (ablation) | INR compression with input encoding · Kodak | rate-distortion (PSNR vs bpp) | Improved |
| add positional encoding to SIREN (ablation) | SIREN INR compression · Kodak | rate-distortion (PSNR vs bpp) | Improved |
| use positional encoding instead of Gaussian encoding (ablation) | INR compression (sine and ReLU) · Kodak | rate-distortion (PSNR vs bpp) | Improved |
| add positional encoding for quantization robustness (ablation) | SIREN basic model · Kodak | quantization error in PSNR | Improved |
Closest papers by meaning
- SINR: Sparsity Driven Compressed Implicit Neural Representations (CVPR 2025 · similarity 0.89)
- NVRC: Neural Video Representation Compression (NeurIPS 2024 · similarity 0.88)
- Implicit Representations via Operator Learning (ICML 2024 · similarity 0.87)
- VI^3NR: Variance Informed Initialization for Implicit Neural Representations (CVPR 2025 · similarity 0.87)
- End-to-End Implicit Neural Representations for Classification (CVPR 2025 · similarity 0.87)
- Grids Often Outperform Implicit Neural Representation at Compressing Dense Signals (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.
