Research map / Video / Video segmentation and tracking
Low-Light Image Enhancement Using Event-Based Illumination Estimation
Lei Sun, Yuhan Bao, Jiajun Zhai, Jingyun Liang, Yulun Zhang, Kaiwei Wang, Danda Pani Paudel, Luc Van Gool
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 temporal-mapping events as input (ablation) | Retinex LLIE pipeline · LOL-v1 | PSNR: 16.98 -> 26.96 | Improved |
| add the T2I module (ablation) | event LLIE pipeline · LOL-v1 | PSNR: 26.96 -> 27.25 | Improved |
| add low-light degradation model (LLDM) to denoiser training (ablation) | T2I module · LOL-v1 | PSNR: 27.25 -> 27.83 | Improved |
| add LLDM when cross-attention fusion is used (ablation) | full model · LOL-v1 | PSNR: 28.45 -> 28.60 | Improved |
| add illumination via simple add fusion in reflectance enhancement (ablation) | event LLIE with LLDM · LOL-v1 | PSNR: 27.83 -> 28.13 | Improved |
| use cross-modal attention instead of add fusion (ablation) | IRE module · LOL-v1 | PSNR: 28.13 -> 28.60 | Improved |
| use cross-modal attention instead of concatenation fusion (ablation) | IRE module · LOL-v1 | PSNR: 28.01 -> 28.60 | Improved |
| use cross-modal attention instead of multiply fusion (ablation) | IRE module · LOL-v1 | PSNR: 27.98 -> 28.60 | Improved |
| use temporal-mapping events for illumination instead of image-only Retinex | Retinexformer vs ours · LOL-v1 | PSNR: 25.16 -> 28.60 | Improved |
| use temporal-mapping events instead of motion events | EvLight vs ours · SDSD-in | PSNR: 28.52 -> 33.65 | Improved |
| use temporal-mapping events instead of motion events | EvLight vs ours · Real-world dataset (60 scenes, exposure-fusion reference) | PIQE: 20.45 -> 9.41 | Improved |
| add motion events to image-only LLIE (EvLight vs best image-only) | EvLight · LOL/SDSD | SSIM: 0.017 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- EvLight — EvLight++: Low-Light Video Enhancement With an Event Camera: A Large-Scale Real-World Dataset, Novel Method, and More (TPAMI 2025) · better on 6 of 6 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
- Towards Robust Event-guided Low-Light Image Enhancement: A Large-Scale Real-World Event-Image Dataset and Novel Approach (CVPR 2024 · similarity 0.92)
- Event-Illumination Collaborative Low-light Image Enhancement with a High-resolution Real-world Dataset (CVPR 2026 · similarity 0.91)
- Low-Light Video Enhancement with Synthetic Event Guidance (AAAI 2023 · similarity 0.91)
- SEE: See Everything Every Time - Adaptive Brightness Adjustment for Broad Light Range Images via Events (IJCV 2026 · similarity 0.90)
- Coherent Event Guided Low-Light Video Enhancement (ICCV 2023 · similarity 0.90)
- Seeing Motion at Nighttime with an Event Camera (CVPR 2024 · 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.
