Research map / Video / Video segmentation and tracking
Seeing Motion in the Dark
Chen Chen, Qifeng Chen, Minh N. Do, Vladlen Koltun
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 siamese self-consistency loss training (ablation) | Ours · DRV static test videos | PSNR (dB): 27.66 -> 28.26 | Improved |
| add VBM4D temporal preprocessing (ablation) | Ours · DRV static test videos | PSNR (dB): 27.62 -> 28.26 | Improved |
| add siamese loss and VBM4D together (ablation) | Ours · DRV static test videos | PSNR (dB): 27.26 -> 28.26 | Improved |
| add siamese self-consistency loss training (ablation) | Ours · DRV static test videos | TPSNR (dB): 37.76 -> 38.31 | Improved |
| add VBM4D temporal preprocessing (ablation) | Ours · DRV static test videos | TPSNR (dB): 34.64 -> 38.31 | Improved |
| add VBM4D preprocessing | SID · DRV static test videos | PSNR (dB): 27.32 -> 27.69 | Improved |
| add VBM4D preprocessing | SID · DRV static test videos | TPSNR (dB): 33.72 -> 37.05 | Improved |
| use end-to-end raw-to-sRGB network instead of denoise then Rawpy (KPN) | Ours vs KPN+Rawpy · DRV static test videos | PSNR (dB): 18.81 -> 28.26 | Improved |
| use our method instead of SID | Ours vs SID · DRV static test videos | PSNR (dB): 27.69 -> 28.26 | Improved |
| use our method instead of SID | Ours vs SID · DRV static test videos | TMAE: 1.05 -> 0.89 | Improved |
| increase self-consistency weight lambda (ablation) | Ours · DRV static test videos | PSNR vs temporal error | Mixed |
| use our method instead of SID in perceptual test | Ours vs SID · Perceptual A/B test, 10 dynamic videos | preference rate (%): 84.12% | Improved |
Closest papers by meaning
- Learning to See in the Dark (CVPR 2018 · similarity 0.90)
- Learning to See Moving Objects in the Dark (ICCV 2019 · similarity 0.88)
- Efficient Real-Time Raw-to-Raw Denoising for Extreme Low-Light Ultra HD Video on Mobile Devices (CVPR 2026 · similarity 0.88)
- Low Light Video Enhancement using Synthetic Data Produced with an Intermediate Domain Mapping (ECCV 2020 · similarity 0.86)
- Binarized Low-light Raw Video Enhancement (CVPR 2024 · similarity 0.86)
- Seeing Dark Videos via Self-Learned Bottleneck Neural Representation (AAAI 2024 · similarity 0.84)
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.
