Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
CVPR 2025 DarkIR: Robust Low-Light Image Restoration - State of the art low light deblurring. NTIRE 2025 Best Method. [Official PyTorch Implementation]
| Date | Stars |
|---|---|
| 2026-07-30 | 352 |
| 2026-08-11 | 353 |
| 2026-08-18 | 355 |
| 2026-08-19 | 356 |
| 2026-08-22 | 357 |
| 2026-08-23 | 358 |
| 2026-08-27 | 359 |
| 2026-09-03 | 360 |
| 2026-09-10 | 361 |
| 2026-09-17 | 362 |
| 2026-09-20 | 362 |
| 2026-09-21 | 363 |
| 2026-10-02 | 363 |
| 2026-10-04 | 364 |
| 2026-10-09 | 365 |
| 2026-10-12 | 367 |
| 2026-10-14 | 366 |
| 2026-10-19 | 367 |
| 2026-10-24 | 367 |
| 2026-10-25 | 368 |
| 2026-10-26 | 367 |
| 2026-10-27 | 368 |
| 2026-10-29 | 369 |
| 2026-11-01 | 370 |
| 2026-11-02 | 371 |
| 2026-11-04 | 371 |
Today
— stars today
This week
+3 stars this week
This month
+7 stars this month
Momentum
0.0
growth rate 0.81%/day
# [CVPR 2025] DarkIR: Robust Low-Light Image Restoration [](https://huggingface.co/spaces/Cidaut/DarkIR) [](https://arxiv.org/abs/2412.13443) **[Daniel Feijoo](https://scholar.google.com/citations?hl=en&user=hqbPn4YAAAAJ), [Juan C. Benito](https://scholar.google.com/citations?hl=en&user=f186MIUAAAAJ), [Alvaro Garcia](https://scholar.google.com/citations?hl=en&user=c6SJPnMAAAAJ), [Marcos V. Conde](https://scholar.google.com/citations?user=NtB1kjYAAAAJ&hl=en)** (CIDAUT AI and University of Wuerzburg) 🚀 The model was presented at CVPR 2025, thanks for your support. Try the model for free in 🤗 [HuggingFace Spaces: DarkIR](https://huggingface.co/spaces/Cidaut/DarkIR), download [model weights/checkpoint](https://cidautes-my.sharepoint.com/:f:/g/personal/alvgar_cidaut_es/Epntbl4SucFNpeIT_jyYZ-cB9BamMbacbyq_svrkMCpShA?e=XB9YBB) and [HF checkpoint](https://huggingface.co/Cidaut/DarkIR/). **TLDR.** In low-light conditions, you have noise and blur in the images, yet, previous methods cannot tackle dark noisy images and dark blurry using a single model. We propose the first approach for all-in-one low-light restoration including illumination, noisy and blur enhancement. *We evaluate our model on LOLBlur, RealLOLBlur, LOL, LOLv2 and LSRW. Follow this repo to receive updates :)* 🔥 [NEWS 2025] DarkIR was a top solution in 3 NTIRE 2025 challenges! - "NTIRE 2024 challenge on low light image enhancement" - "NTIRE 2025 challenge on efficient burst hdr and restoration" - "NTIRE 2025 challenge on day and night raindrop removal for dual-focused images" <details> <summary> <b> ABSTRACT </b> </summary> >Photography during night or in dark conditions typically suffers from noise, low light and blurring issues due to the dim environment and the common use of long exposure. Although Deblurring and Low-light Image Enhancement (LLIE) are related under these conditions, most approaches in image restoration solve these tasks separately. In this paper, we present an efficient and robust neural network for multi-task low-light image restoration. Instead of following the current tendency of Transformer-based models, we propose new attention mechanisms to enhance the receptive field of efficient CNNs. Our method reduces the computational costs in terms of parameters and MAC operations compared to previous methods. Our model, DarkIR, achieves new state-of-the-art results on the popular LOLBlur, LOLv2 and Real-LOLBlur datasets, being able to generalize on real-world night and dark images. </details> | <img src="assets/teaser/0085_low.png" alt="add" width="450"> | <img src="assets/teaser/0085_retinexformer.png" alt="add" width="450"> | <img src="assets/teaser/0085_darkir.png" alt="add" width="450"> | |:-------------------------:|:-------------------------:|:-------------------------:| | Low-light w/ blur | RetinexFormer | **DarkIR** (ours) | | <img src="assets/teaser/low00747.png" alt="add" width="450"> | <img src="assets/teaser/low00747_lednet.png" alt="add" width="450"> | <img src="assets/teaser/low00747_darkir.png" alt="add" width="450"> | | Low-light w/o blur | LEDNet | **DarkIR** (ours) | ## Network Architecture  ## Dependencies and Installation - Python == 3.10.12 - PyTorch == 2.5.1 - CUDA == 12.4 - Other required packages in `requirements.txt` ``` # git clone this repository git clone https://github.com/Fundacion-Cidaut/DarkIR.git cd DarkIR # create python environment python3 -m venv venv_DarkIR source venv_DarkIR/bin/activate # install python dependencies pip install -r requirements.txt ``` ## Datasets The datasets used for training and/or evaluation are: |Dataset | Sets of images | Source | | -----------| :---------------:|------| |LOL-Blur | 10200 training pairs / 1
Excerpt of 9,701 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:e334e3ee91a00729, topic:deep-learning
matched fp:e334e3ee91a00729, topic:computer-vision