This is an unofficial pytorch implementation of AHDRNet

• python 3.8
• pytorch 1.11.0
• Use the requirements.txt file to install the necessary packages
  • pip install -r requirements.txt

• Download here
• Organize the dataset file as follows, and set the dataset path as the Main Path in args_file.py
• There are only 10 groups of test data under the original EXTRA folder, you can add the five groups of data under the PAPER folder to the EXTRA folder, rename it 011 - 015, and note that there are only five files under the folder to check each group of data: three tif format pictures, one hdr format picture, and one txt folder

Main Path
├─Test
│  ├─EXTRA
│  │  ├─001
|  |  | ├── 1.tif
|  |  | ├── 2.tif
|  |  | ├── 3.tif
|  |  | ├── exposure.txt
|  |  | └── HDRImg.hdr (optional)
│  │  ├─002
│  │  ├─003
│  │  ├─004
│  │  ├─005
│  │  ├─006
│  │  ├─007
│  │  ├─008
│  │  ├─009
│  │  ├─010
│  │  ├─011
│  │  ├─012
│  │  ├─013
│  │  ├─014
│  │  └─015
│  └─PAPER
│      ├─BarbequeDay
│      ├─LadySitting
│      ├─ManStanding
│      ├─PeopleStanding
│      └─PeopleTalking
└─Training
    ├─001
    ├─002
    ├─003
    ├─004
    └─...

1. Open args_file.py file to set training related parameters
2. Run train.py to Train
3. Follow the prompts to open the tensorboard to observe the validation results during training, the loss function changes

1. Open fusion_HDR.py file, Set weight path, dataset path, path to save fusion results
2. Run fusion_HDR.py to get HDR image.

• https://github.com/qingsenyangit/AHDRNet