Research map / Video / Video segmentation and tracking
Ensemble Learning Priors Driven Deep Unfolding for Scalable Video Snapshot Compressive Imaging
Chengshuai Yang, Shiyu Zhang, Xin Yuan
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 ensemble priors in the last stage (13-6 vs 13-1) (ablation) | ELP-Unfolding (128 channels, no connection) · SCI benchmark ablation (128 channels, no connection) | PSNR (dB): 31.75 -> 33.46 | Improved |
| add ensemble priors (6 priors vs 1 prior, 13 stages) (ablation) | ELP-Unfolding full model · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 35.07 -> 35.41 | Improved |
| add dense connection to the 13-6 model (ablation) | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 34.77 -> 35.41 | Improved |
| add dense connection to the 7-1 model (ablation) | ELP-Unfolding 7 stages, 1 prior · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 34.23 -> 34.85 | Improved |
| add ensemble only in later stages instead of ensembling every stage (ablation) | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 34.97 -> 35.41 | Improved |
| add ensemble priors vs single prior in all stages (ablation) | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 35.09 -> 35.41 | Improved |
| use full DAVIS2017 training set instead of part (ablation) | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 34.73 -> 35.41 | Improved |
| increase number of stages from 1 to 13 (ablation) | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 31.21 -> 35.07 | Improved |
| increase stages from 11 to 13 (1 prior) (ablation) | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 35.11 -> 35.07 | No clear effect |
| increase priors in last stage from 6 to 8 (ablation) | ELP-Unfolding 13 stages · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 35.41 -> 35.34 | Hurt |
| replace Dense3D-Unfolding 3D-CNN with 2D-CNN ensemble unfolding | ELP-Unfolding · SCI benchmark (6 scenes, B=8) average | PSNR (dB): 35.26 -> 35.41 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Meta-SCI — MetaSCI: Scalable and Adaptive Reconstruction for Video Compressive Sensing (CVPR 2021) · better on 2 of 2 shared results
- BIRNAT — BIRNAT: Bidirectional Recurrent Neural Networks with Adversarial Training for Video Snapshot Compressive Imaging (ECCV 2020) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Dense Deep Unfolding Network With 3D-CNN Prior for Snapshot Compressive Imaging (ICCV 2021 · similarity 0.91)
- Untrained Neural Nets for Snapshot Compressive Imaging: Theory and Algorithms (NeurIPS 2024 · similarity 0.91)
- Unfolding Framework with Prior of Convolution-Transformer Mixture and Uncertainty Estimation for Video Snapshot Compressive Imaging (ICCV 2023 · similarity 0.91)
- Memory-Efficient Network for Large-Scale Video Compressive Sensing (CVPR 2021 · similarity 0.90)
- Dual Prior Unfolding for Snapshot Compressive Imaging (CVPR 2024 · similarity 0.87)
- Using Powerful Prior Knowledge of Diffusion Model in Deep Unfolding Networks for Image Compressive Sensing (CVPR 2025 · similarity 0.86)
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.
