PDDA

KO | EN

gitlite — search

#javascript #python #react #typescript #llm #golang #machine-learning #deep-learning #macos #awesome #cli #pytorch #mcp #openai

(★ 9)

(IEEE TIP 2026) PDDA: Prompt-driven Domain Adaptation for Real-world Image Dehazing

PDDA Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h1>PDDA: Prompt-driven Domain Adaptation for Real-world Image Dehazing</h1>

<div>
    Yan Zhang</a>&emsp;
    Yuxin Feng</a>&emsp;
    Xin Li</a>&emsp;
    Fan Zhou</a>&emsp;
    Zhuo Su</a>
</div>

<div>
    <h4 align="center">
        <a href="https://pan.baidu.com/s/1xO6ddmlDO_iom7uGxACLqA?pwd=0519" target='_blank'>[Weight]</a>
    </h4>
</div>

</div>

## :mega: Updates
- **2025.04.03**: :fire::fire::fire: Training codes, and initial checkpoints are publicly available now.
- **2026.04.03**: :tada::tada::tada: Accepted by ***IEEE TIP***.
## :pencil2: Results

## :desktop_computer: Environment

Step 1. Clone this repo:

```
git clone https://github.com/YanZhang-zy/PDDA.git
cd PDDA/
```

Step 2. Create a new conda environment and install dependencies:

```
conda create -n PDDA python=3.10
conda activate PDDA
conda install pytorch==2.2.0 torchvision==0.17.0 torchaudio==2.2.0 pytorch-cuda=11.8 -c pytorch -c nvidia
pip install -r requirements.txt
```

##  :book: Data Preparation

Step 1. Download the haze dataset from websites or papers.

Step 2. Make sure the file structure is consistent with the following:

```
dataset/
├── HAZE4K
│   ├── test
│   |   ├── clear
│   |   │   ├── 1.png
│   |   │   └── 2.png
│   |   │   └── ...
│   |   └── hazy
│   |       ├── 1_0.89_1.56.png
│   |       └── 2_0.93_1.66.png
│   |       └── ...
│   └── train
│       ├── clear
│       │   ├── 1.png
│       │   └── 2.png
│       │   └── ...
│       └── hazy
│           ├── 1_0.68_0.66.png
│           └── 2_0.59_1.95.png
│           └── ...
├── REAL_FOGGY
│   ├── 1.png
│   └── 2.png
│   └── 3.png
│   └── ...
```

The datasets can be downloaded at
+ [RESIDE](https://sites.google.com/view/reside-dehaze-datasets/reside-v0)
+ [HAZE4K](https://pan.baidu.com/s/19stkJ3aaF8WgHK2FBytnZA?pwd=0411)
+ [REAL_FOGGY](https://pan.baidu.com/s/1GS9qkwcBcKB411pdSwFcDg?pwd=0519)

## :train: Pre-trained Haze Prompt Pairs Training

In our experiment, we randomly select 4000 haze images from URHI training dataset and 2000 clear images from RESIDE-Outdoor testing dataset as the unpaired training data. We provide the training data we use at [[BaiduPan](https://pan.baidu.com/s/1G8yiq6CWRzVHEZKhu-SD7A?pwd=0314)] for your reference. For the specific training code, please refer to [CLIP-LIT](https://github.com/ZhexinLiang/CLIP-LIT.git).

## :hotsprings: Model Training
Step 1. Download the pre-trained Res2Net model weight from [[BaiduPan](https://pan.baidu.com/s/1xO6ddmlDO_iom7uGxACLqA?pwd=0519)].

Step 2. Make sure the file structure is consistent with the following:
```
pre_model/
└── res2net101_v1b_26w_4s-0812c246.pth
clip_model/
├── haze_prompt.pth
├── RN101.pt
└── ViT-B-32.pt
```

Step 3. Run the following script to train PDDA from scratch:
```
python NBL.py
python BL.py
```

## :taxi: Model Testing
Step 1. Download the pre-trained model weights from [[BaiduPan](https://pan.baidu.com/s/1xO6ddmlDO_iom7uGxACLqA?pwd=0519)].

Step 2. Make sure the file structure is consistent with the following:
```
weight/
├── Pre.pth
└── BL.pth
```

Step 3. Run the following script to test PDDA:
```
python Eval.py
```

PDDA

PDDA Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

PDDA

PDDA Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README