ZSDECNet : A Zero-Shot Deep Learning Framework for Image Exposure Correction

This is the official PyTorch code for our paper : ZSDECNet: A Zero-Shot Deep Learning Framework for Image Exposure Correction

Wenchao Li, Shuyuan Wen,Jinhao Zhua, Qiaofeng Oub, Yanchun Guob, Jiabao Chen and Bangshu Xiong

Keywords: Exposure correction, Zero-shot, Illumination estimation, Multi-exposure fusion.

This repository contains the code and resources for the paper:

code: Will be open source after the paper is accepted.

resources: includes the DSN night street view dataset and test results on all datasets.

Setup

Create a new conda environment

  conda create -n ZSDECNet python=3.7
  conda activate ZSDECNet

Install Pytroch

  # CUDA  
  conda install pytorch==1.8.1 torchvision==0.9.1 torchaudio==0.8.1 cudatoolkit=10.2 -c pytorch

DSN Night street scene, including underexposure and overexposure. Can be downloaded from Baidu Cloud and Google Drive.

Contains the visual comparison results of all methods in the paper on 7 datasets. Can be downloaded from Baidu Cloud and and Google Drive.

If you have any problem with the released code, please do not hesitate to contact me by email (wenchaoli@buaa.edu.cn)

xxx