Research map / Image generation / Image restoration
Consensus Convolutional Sparse Coding
Biswarup Choudhury, Robin Swanson, Felix Heide, Gordon Wetzstein, Wolfgang Heidrich
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 |
|---|---|---|---|
| solve CSC as a consensus problem over blocks (CCSC, 10 workers) instead of global CSC | convolutional sparse coding · 2D dataset of 500 images | runtime (sec): 1530.71 sec -> 82.69 sec | Improved |
| increase parallel workers from 10 to 50 in CCSC (ablation) | CCSC · 2D dataset of 500 images | runtime (sec): 82.69 sec -> 28.57 sec | Improved |
| use CCSC instead of global CSC on 1000 images | convolutional sparse coding · 2D dataset of 1000 images | feasibility: Out of Memory -> 387.68 sec | Improved |
| use CCSC (5 workers) at 100 images | convolutional sparse coding · 2D dataset of 100 images | runtime (sec): 203.56 sec -> 35.35 sec | Improved |
| split the CSC problem into consensus subproblems (ablation) | CCSC · iteration count | iterations needed | Hurt |
| use CCSC large-scale ImageNet dictionary for inpainting versus conventional CSC | CCSC · 2D inpainting, Wind Mill | PSNR (dB): 33.49 -> 35.13 | Improved |
| use CCSC dictionary for Poisson deconvolution versus conventional CSC | CCSC · 2D Poisson deconvolution, Porsche | PSNR (dB): 26.21 -> 27.11 | Improved |
| use CCSC for multispectral demosaicing versus IID | CCSC · 3D multispectral demosaicing, Clay | PSNR (dB): 14.25 -> 29.74 | Improved |
| use CCSC for light field view synthesis versus Kalantari | CCSC · Light field view synthesis, Cars | PSNR (dB): 21.41 -> 27.57 | Improved |
| train dictionary on 5000 ImageNet images instead of a very small fruit dataset (ablation) | CCSC · reconstruction | reconstruction quality | Improved |
| learn multispectral dictionary with shared coefficients across channels instead of solving each channel individually with 2D CSC (ablation) | CCSC · multispectral data | reconstruction quality | Improved |
Closest papers by meaning
- High Order Tensor Formulation for Convolutional Sparse Coding (ICCV 2017 · similarity 0.91)
- Rethinking the CSC Model for Natural Images (NeurIPS 2019 · similarity 0.91)
- Convolutional Dictionary Learning via Local Processing (ICCV 2017 · similarity 0.91)
- SparseFormer: Sparse Visual Recognition via Limited Latent Tokens (ICLR 2024 · similarity 0.86)
- CPP-Net: Embracing Multi-Scale Feature Fusion into Deep Unfolding CP-PPA Network for Compressive Sensing (CVPR 2024 · similarity 0.85)
- Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation (ICML 2025 · 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.
