Awesome Low-light Enhancement

If you have any problems, suggestions or improvements, please submit the issue or PR.

Paper with Code

Survey on Arxiv (2022-12)

---

Contents

• Datasets
• Papers

Datasets

• VV, LIME, NPE-series, DICM, MEF

  • only low-light images without corresponding high-light ground truth
  • Download (Thanks to baidut for collection)

• SID

  • Homepage
  • Download: Sony part + Fuji part or Baidu Drive

• MIT-Adobe 5K

  • Homepage
  • Download (use only the output by Expert C)

• ExDARK

  • only low-light image without corresponding high-light ground truth
  • Homepage
  • Download

• LOL

  • Homepage
  • Download: Google Drive or Baidu Pan (Code: acp3)

• LOL-Real (LOL v2):

  • Google Drive or Baidu Pan (Code: l9xm)
  • both V1 & V2 can be found here

• SMID-DRV (dark raw video dataset)

  • Homepage

• SICE

  • Homepage
  • Google Drive: part1 + part2
  • Baudu Yun: part1 + part2 + part2 Label

• DeepUPE (not public)

• ELD

  • Homepage
  • Download: Google Drive or Baidu Pan (Code: 0lby)

• LSRW

  • Paper (R2RNet): Arxiv or J Vis Commun Image R
  • Homepage (github project)
  • Download: Baidu Pan (Code: wmrr)

• DARK FACE

  • joint with face detection
  • Homepage
  • Google Drive: train&val + test
  • Baidu Pan: train&val (Code: babu) + test (Code: 429h)

• ACDC

  • 1006 nighttime images, joint with semantic segmentation
  • Benchmark
  • Overview & Download

• SDSD

  • video dataset for supervised methods
  • Homepage
  • Download: Google Drive or Baidu Pan (Code: zcrb)

• Dark Zurich

  • contains corresponding images of the same driving scenes at daytime, twilight and nighttime.
  • homepage

• Real-LOL-Blur

  • joint with deblur
  • Homepage
  • Download (Google Drive)

• MSEC

  • each image contains either over- or under-exposure errors
  • homepage
  • Training & Validation & Test

• LCD

  • each image contains both over- and under-exposure errors.
  • homepage
  • download

• GTA5

  • Synthetic GTA5 nighttime fog data [paper]
  • download

Papers

Supervised Method

• [SID] Learning to See in the Dark (CVPR 2018) [paper][code][code-pytorch]

• [Retinex-Net] Deep Retinex Decomposition for Low-Light Enhancement (BMVC 2018) [paper][code][code-pytorch]

• [HDRNet] Deep Bilateral Learning for Real-Time Image Enhancement (SIGGRAPH 2017) [paper][code]

• [DeepUPE] Underexposed Photo Enhancement using Deep Illumination Estimation (CVPR 2019) [paper][code (only test)]

• [CWAN] Color-wise Attention Network for Low-light Image Enhancement [paper]

• [SMID] Seeing Motion in the Dark (ICCV 2019) [homepage][code-tensorflow]

• [MSR-net] MSR-net:Low-light Image Enhancement Using Deep Convolutional Network [paper]

• [GLADNet] GLADNet: Low-Light Enhancement Network with Global Awareness (FG 2018) [paper][code]

• [LIE-GP] Low-light image enhancement using Gaussian Process for features retrieval (SPIC) [paper][code]

• [MBLLEN] MBLLEN: Low-light Image/Video Enhancement Using CNNs (BMVC 2018) [paper][code]

• [LLNet] LLNet: Low-light Image Enhancement with Deep Learning ((PR)) [paper][code]

• [KinD] Kindling the Darkness: A Practical Low-light Image Enhancer (ACM MM 2019) [paper] [code]

• [LL-RefineNet] Deep Refinement Network for Natural Low-Light Image Enhancement in Symmetric Pathways (Symmetry) [paper]

• [LLCNN] LLCNN: A convolutional neural network for low-light image enhancement (VCIP 2017) [paper][code]

• [LLEDHN] Low-Light Image Enhancement via a Deep Hybrid Network (T-IP) [paper][code]

• [SICE] Learning a Deep Single Image Contrast Enhancer from Multi-Exposure Images (T-IP) [paper][code]

• [ALIE] Attention-guided Low-light Image Enhancement [paper] [homepage]

• [DCGANs] Deep Learning for Robust end-to-end Tone Mapping (BMVC 2019) [paper]

• [RJI] Robust Joint Image Reconstruction from Color and Monochrome Cameras (BMVC 2019) [paper]

• [FIDE] Learning to Restore Low-Light Images via Decomposition-and-Enhancement (CVPR 2020) [paper]

• [Pb-NFM] A Physics-based Noise Formation Model for Extreme Low-light Raw Denoising (CVPR 2020, paper for ELD dataset) [paper][Code]

• [DALE] DALE : Dark Region-Aware Low-light Image Enhancement (BMVC 2020) [paper]

• [LLPackNet] Towards Fast and Light-Weight Restoration of Dark Images (BMVC 2020) [[paper] [code]

• [StableLLVE] Learning Temporal Consistency for Low Light Video Enhancement from Single Images (CVPR 2021) [paper][Code]

• [UTVNet] Adaptive Unfolding Total Variation Network for Low-Light Image Enhancement (ICCV 2021) [Paper][Code]

• [SNRA] SNR-Aware Low-light Image Enhancement (CVPR 2022) [paper][code]

• [URetinex] URetinex-Net: Retinex-Based Deep Unfolding Network for Low-Light Image Enhancement (CVPR 2022) [paper]

• [DCC-Net] Deep Color Consistent Network for Low-Light Image Enhancement [paper]

• [LEDNet] LEDNet: Joint Low-light Enhancement and Deblurring in the Dark (ECCV 2022) [paper][code]

• [LCDPNet] Local Color Distributions Prior for Image Enhancement (ECCV 2022) [paper][homepage]

• [IAT] You Only Need 90K Parameters to Adapt Light: a Light Weight Transformer for Image Enhancement and Exposure Correction (BMVC 2022) [project][paper][code]

• [CFA-LLVE] Low Light Video Enhancement by Learning on Static Videos with Cross-Frame Attention (BMVC 2022) [project][paper]

Semi-Superised and Unsupervised Method

• [LIME] LIME: Low-light Image Enhancement via Illumination Map Estimation (T-IP) [paper] [code-matlab][code-numpy]

• [EnlightenGAN] EnlightenGAN: Deep Light Enhancement without Paired Supervision [paper][code]

• [Zero-DCE] Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement (CVPR 2020) [paper][homepage][code]

• [DRBN] From Fidelity to Perceptual Quality: A Semi-Supervised Approach for Low-Light Image Enhancement (CVPR 2020) [Paper][Code]

• [RUAS] Retinex-inspired Unrolling with Cooperative Prior Architecture Search for Low-light Image Enhancement (CVPR 2021) [Paper][Code]

• [SSIE] Self-supervised Image Enhancement Network: Training with Low Light Images Only (Arxiv) [paper][code]

• [SCI] Toward Fast, Flexible, and Robust Low-Light Image Enhancement (CVPR 2022) [paper][Code]

• [LES] Unsupervised Night Image Enhancement: When Layer Decomposition Meets Light-Effects Suppression (ECCV 2022) [paper][code]

• [UDCL-Transformer] Unsupervised Low Light Image Enhancement Transformer Based on Dual Contrastive Learning (BMVC 2022) [project][paper][code]