bozhiliu85/rassdl

Tone mapping high dynamic range images based on region-adaptive self-supervised deep learning

RASSDL

Tone mapping high dynamic range images based on region-adaptive self-supervised deep learning

Zhou, Fei, Guangsen Liao, Jiang Duan, Bozhi Liu, and Guoping Qiu. "Tone mapping high dynamic range images based on region-adaptive self-supervised deep learning." Signal Processing: Image Communication 102 (2022): 116595.

environment

• Python 3.8
• PyTorch

Data Download

We collected HDR images from four databases:

Fairchild, http://rit-mcsl.org/fairchild//HDR.html

HDR-Eye, http://infoscience.epfl.ch/record/203873

Anyhere, http://www.anyhere.com/gward/hdrenc/pages/originals.html

Stanford-HDRI, https://exhibits.stanford.edu/data/catalog/sz929jt3255

Data Preprocessing

run dataset.py, prepare the train data list train.txt and test data list test.txt

Train

run main_train.py

Test

run main_test.py to reproduce the results on the paper

Citation

@article{zhou2022tone, title={Tone mapping high dynamic range images based on region-adaptive self-supervised deep learning}, author={Zhou, Fei and Liao, Guangsen and Duan, Jiang and Liu, Bozhi and Qiu, Guoping}, journal={Signal Processing: Image Communication}, volume={102}, pages={116595}, year={2022}, publisher={Elsevier} }