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.
Code for CVPR 2020 paper "Deep Local Parametric Filters for Image Enhancement"
| Date | Stars |
|---|---|
| 2026-07-24 | 260 |
| 2026-07-25 | 260 |
| 2026-07-28 | 260 |
| 2026-07-30 | 260 |
| 2026-08-06 | 260 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# DeepLPF: Deep Local Parametric Filters for Image Enhancement (CVPR 2020)
[](https://github.com/sjmoran/deeplpf-image-enhancement/actions/workflows/ci.yml)
[](https://arxiv.org/abs/2003.13985)
[](https://github.com/sjmoran/deeplpf-image-enhancement/stargazers)
[Sean Moran](https://sjmoran.github.io/), [Pierre Marza](https://pierremarza.github.io/), [Steven McDonagh](https://smcdonagh.github.io/), [Sarah Parisot](https://parisots.github.io/), [Greg Slabaugh](https://eecs.qmul.ac.uk/~gslabaugh/) — **Huawei Noah's Ark Lab**
[[Paper]](https://arxiv.org/abs/2003.13985) [[Poster]](https://github.com/sjmoran/sjmoran.github.io/blob/main/pdfs/DeepLPF_CVPR20_poster.pdf) [[Video]](https://www.youtube.com/watch?v=Sxach3FM6FY) [[Supplementary]](https://github.com/sjmoran/sjmoran.github.io/blob/7775d1fc39d14baeb6935f6c750f923e1251f491/pdfs/DeepLPF_supplementary.pdf)
Official PyTorch implementation of the CVPR 2020 paper **DeepLPF: Deep Local Parametric Filters for Image Enhancement**. Instead of predicting output pixels directly, DeepLPF regresses the parameters of a small set of spatially localised image filters (cubic, graduated and elliptical) and applies them, giving an interpretable retouching model. On the Adobe-DPE benchmark the bundled pre-trained model reaches **23.90 dB PSNR / 0.911 SSIM**.
<p align="center">
<img src="./images/teaser.png" width="80%"/>
</p>
## Contents
- [Quick start](#quick-start)
- [Results](#results)
- [How it works](#how-it-works)
- [Pre-trained models](#pre-trained-models)
- [Training](#training)
- [Datasets](#datasets)
- [Original (CVPR 2020) code](#original-cvpr-2020-code)
- [Citation](#citation)
- [License](#license)
- [Contributing](#contributing)
- [Errata](#errata)
## Quick start
```bash
git clone https://github.com/sjmoran/deeplpf-image-enhancement.git
cd deeplpf-image-enhancement
pip install -r requirements.txt
# Enhance the bundled example images with the pre-trained Adobe-DPE model.
# Results (with PSNR/SSIM in the filenames) appear in a timestamped log_* directory.
python3 main.py \
--inference_img_list_path=./adobe5k_dpe/images_inference.txt \
--inference_img_dirpath=./adobe5k_dpe/ \
--checkpoint_filepath=./pretrained_models/adobe_dpe/deeplpf_validpsnr_23.378_validloss_0.033_testpsnr_23.904_testloss_0.031_epoch_424_model.pt
```
The code picks the best available device automatically: a CUDA GPU, Apple Silicon (MPS), or CPU. No configuration is needed.
## Results
The bundled `adobe_dpe` checkpoint (epoch 424) on the Adobe-DPE test set:
| Dataset | PSNR | SSIM |
|---|---|---|
| Adobe-DPE | 23.90 dB | 0.911 |
Input → expert-retouched label → DeepLPF output:
<table>
<tr><th>Input</th><th>Label</th><th>Ours (DeepLPF)</th></tr>
<tr>
<td><img src="./adobe5k_dpe/deeplpf_example_test_input/a4576-DSC_0217_input.png"/></td>
<td><img src="./adobe5k_dpe/deeplpf_example_test_output/a4576-DSC_0217_gt.png"/></td>
<td><img src="./adobe5k_dpe/deeplpf_example_test_inference/a4576-DSC_0217_TEST_425_9_PSNR_34.596_SSIM_0.980.jpg"/></td>
</tr>
<tr>
<td><img src="./adobe5k_dpe/deeplpf_example_test_input/a4742-Duggan_090331_6517_input.png"/></td>
<td><img src="./adobe5k_dpe/deeplpf_example_test_output/a4742-Duggan_090331_6517_output.png"/></td>
<td><img src="./adobe5k_dpe/deeplpf_example_test_inference/a4742-Duggan_090331_6517_TEST_425_1_PSNR_29.825_SSIM_0.984.jpg"/></td>
</tr>
<tr>
<td><img src="./adobe5k_dpe/deeplpf_example_test_input/a4869-09-05-19-at-19h05m38s-_MG_9563.png"/></td>
<td><img src="./adobe5k_dpe/deeplpf_example_test_output/a4869-09-05-19-at-19h05m38s-_MG_9563.png"/></td>
<td><img src="./adobe5k_dpe/deeplpf_example_test_inferenceExcerpt of 11,984 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:90d9fc70028fa428, topic:deep-learning
matched fp:90d9fc70028fa428, topic:computer-vision