/SAMHDR

Single-Image HDR Reconstruction Assisted Ghost Suppression and Detail Preservation Network for Multi-Exposure HDR Imaging (Accepted by IEEE TCI, 2024)

Primary LanguagePython

SAMHDR

Python >=3.8 PyTorch >=1.10

Single-Image HDR Reconstruction Assisted Ghost Suppression and Detail Preservation Network for Multi-Exposure HDR Imaging

The official repository for [Single-Image HDR Reconstruction Assisted Ghost Suppression and Detail Preservation Network for Multi-Exposure HDR Imaging].

Pipeline

framework

Requirements

Installation

we use /torch >=1.10 / 24G  RTX3090 for training and evaluation.

Prepare Datasets

mkdir data

Download the HDR datasets Kalantari's data and Hu's data.

Then unzip them and rename them under the directory like.

data
├── Training
│   └── 
│   └── 
│   └── 
├── Test
│   └── 
│   └── 
│   └──

To generate the enhanced stop image (ESI) as required in the paper.

cd dataset
python data_argument.py

Train

We utilize 1 RTX3090 GPU for training.

python train.py

Evaluation

python test.py

Contact

If you have any questions, please feel free to contact me.(yangtongxue123_km@163.com).

Citation