Research map / Architectures and efficiency / RNNs and normalization
Full Resolution Image Compression With Recurrent Neural Networks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train on hard-to-compress High Entropy patches rather than the 32×32 dataset (ablation) | Image compression networks · Kodak | Rate-distortion AUC | Helped |
| Add entropy coding to the image codes | Recurrent image compression models · Kodak | Bit-rate: 5% at 2 bpp, and 32% at 0.25 bpp | Helped |
| Add entropy coding to the Baseline LSTM codes | Baseline LSTM [17] · Large images | Bit-rate saving: 25% at 2 bpp to 57% at 0.25 bpp | Helped |
| Use one-shot rather than additive reconstruction (ablation) | LSTM compression model · Kodak, trained on 32×32 dataset | MS-SSIM AUC: 1.8041 -> 1.8062 | Helped |
| Use residual scaling rather than one-shot reconstruction (ablation) | GRU compression model · Kodak, trained on 32×32 dataset | MS-SSIM AUC: 1.8098 -> 1.7948 | Hurt |
| Use residual GRU rather than GRU (ablation) | One-shot compression model · Kodak, trained on High Entropy dataset | PSNR-HVS AUC: 53.07 -> 53.19 | Helped |
| Use LSTM rather than residual GRU (ablation) | One-shot compression model · Kodak, trained on High Entropy dataset | PSNR-HVS AUC: 53.19 -> 48.86 | Hurt |
| Use residual GRU rather than LSTM (ablation) | One-shot compression model · Kodak, trained on High Entropy dataset | MS-SSIM AUC: 1.8166 -> 1.8119 | Hurt |
Closest papers by meaning
- Improved Lossy Image Compression With Priming and Spatially Adaptive Bit Rates for Recurrent Networks (CVPR 2018 · similarity 0.92)
- End-to-end Optimized Image Compression (ICLR 2017 · similarity 0.89)
- Towards Backward-Compatible Continual Learning of Image Compression (CVPR 2024 · similarity 0.87)
- Robustly overfitting latents for flexible neural image compression (NeurIPS 2024 · similarity 0.87)
- GLIC: General Format Learned Image Compression (AAAI 2025 · similarity 0.87)
- Fitted Neural Lossless Image Compression (CVPR 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.