Research map / Video / Video segmentation and tracking
BIRNAT: Bidirectional Recurrent Neural Networks with Adversarial Training for Video Snapshot Compressive Imaging
Ziheng Cheng, Ruiying Lu, Zhengjue Wang, Hao Zhang, Bo Chen, Ziyi Meng, 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 self-attention to the first-frame CNN (ablation) | BIRNAT · Video SCI simulation (6 datasets, average) | PSNR: 33.11 -> 33.31 | Improved |
| add adversarial training (ablation) | BIRNAT · Video SCI simulation (6 datasets, average) | PSNR: 33.18 -> 33.31 | Improved |
| add adversarial training (ablation) | BIRNAT · Video SCI simulation (6 datasets, average) | SSIM: 0.9494 -> 0.9507 | Improved |
| add self-attention to the first-frame CNN (ablation) | BIRNAT · Video SCI simulation (6 datasets, average) | SSIM: 0.9467 -> 0.9507 | Improved |
| add self-attention on top of a model without adversarial training (ablation) | BIRNAT w/o AT · Video SCI simulation (6 datasets, average) | PSNR: 32.98 -> 33.18 | Improved |
| add backward RNN (ablation) | BIRNAT w/o SA&AT · Video SCI simulation (6 datasets, average) | PSNR: 31.86 -> 32.98 | Improved |
| add backward RNN (ablation) | BIRNAT w/o SA&AT · Video SCI simulation (6 datasets, average) | SSIM: 0.9343 -> 0.9389 | Improved |
| add backward RNN (ablation) | BIRNAT w/o SA&AT · Video SCI simulation (6 datasets, average) | running time (s): 0.0856 -> 0.1489 | Hurt |
| BIRNAT vs DeSCI on complex-background Aerial | BIRNAT · Aerial | PSNR: 25.33 -> 28.99 | Improved |
| BIRNAT vs DeSCI on Kobe (fast object motion) | BIRNAT · Kobe | PSNR: 33.25 -> 32.71 | Hurt |
| BIRNAT vs DeSCI on Drop | BIRNAT · Drop | PSNR: 43.22 -> 42.28 | Hurt |
| BIRNAT vs U-net in average PSNR | BIRNAT · Video SCI simulation (6 datasets, average) | PSNR: 29.45 -> 33.31 | Improved |
Later papers that compared with this method
9 later accepted papers report BIRNAT as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- 3One2: One-Step Regression plus One-Step Diffusion for One-Hot Modulation in Dual-Path Video Snapshot Compressive Imaging (AAAI 2026 · better on 1 of 1 shared result)
- Motion-Aware Dynamic Graph Neural Network for Video Compressive Sensing (TPAMI 2024 · better on 2 of 2 shared results)
- A Simple Low-bit Quantization Framework for Video Snapshot Compressive Imaging (ECCV 2024 · better on 1 of 1 shared result)
- Hierarchical Separable Video Transformer for Snapshot Compressive Imaging (ECCV 2024 · better on 1 of 1 shared result)
- Unfolding Framework with Prior of Convolution-Transformer Mixture and Uncertainty Estimation for Video Snapshot Compressive Imaging (ICCV 2023 · better on 1 of 1 shared result)
- Ensemble Learning Priors Driven Deep Unfolding for Scalable Video Snapshot Compressive Imaging (ECCV 2022 · better on 2 of 2 shared results)
- Dual-view Snapshot Compressive Imaging via Optical Flow Aided Recurrent Neural Network (IJCV 2021 · better on 2 of 4 shared results)
- Memory-Efficient Network for Large-Scale Video Compressive Sensing (CVPR 2021 · better on 2 of 3 shared results)
1 more not listed.
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
- Recurrent Neural Networks for Snapshot Compressive Imaging (TPAMI 2022 · similarity 0.97)
- Dual-view Snapshot Compressive Imaging via Optical Flow Aided Recurrent Neural Network (IJCV 2021 · similarity 0.90)
- Memory-Efficient Network for Large-Scale Video Compressive Sensing (CVPR 2021 · similarity 0.89)
- Untrained Neural Nets for Snapshot Compressive Imaging: Theory and Algorithms (NeurIPS 2024 · similarity 0.89)
- MetaSCI: Scalable and Adaptive Reconstruction for Video Compressive Sensing (CVPR 2021 · similarity 0.88)
- VideoOneNet: Bidirectional Convolutional Recurrent OneNet with Trainable Data Steps for Video Processing (ICML 2020 · similarity 0.88)
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.
