deephdr's Introduction

ECCV'18: Deep High Dynamic Range Imaging with Large Foreground Motions

This is the implementation for Deep High Dynamic Range Imaging with Large Foreground Motions, Shangzhe Wu, Jiarui Xu, Yu-Wing Tai, Chi-Keung Tang, in ECCV, 2018. More results can be found on our project page.

Get Started

Prerequisites

Python 3.5
Tensorflow 1.4.0
OpenCV 3.4 (from conda-forge)
Photomatix for tonemapping

Setup

Clone this repo:

git clone https://github.com/elliottwu/DeepHDR.git
cd DeepHDR

Download pretrained model: (~60MB)

sh download_pretrained.sh

Demo

sh test.sh

Tonemapping (post-processing)

Generated HDR images are in .hdr format, which may not be properly displayed in your image viewer directly. You may use Photomatix for tonemapping:

Download Photomatix free trial, which won't expire.
Load the generated .hdr file in Photomatix.
Adjust the parameter settings. You may refer to pre-defined styles, such as Detailed and Painterly2.
Save your final image in .tif or .jpg.

Train

Download Kalantari's dataset: (~8GB)

cd dataset
sh download_dataset.sh
cd ..

Prepare TFRecord: (this takes ~10 minutes)

cd dataset
python convert_to_tfrecord.py
cd ..

Start training:

sh train.sh

To monitor training using Tensorboard, copy the following to your terminal and open localhost:8888 in your browser

tensorboard --logdir=logs --port=8888

Citation

@article{szwuDeepHDR,
  author    = {Shangzhe Wu and Jiarui Xu and Yu{-}Wing Tai and Chi{-}Keung Tang},
  title     = {Deep High Dynamic Range Imaging with Large Foreground Motions},
  journal   = {CoRR},
  volume    = {abs/1711.08937},
  year      = {2017},
  url       = {http://arxiv.org/abs/1711.08937},
}

Recommend Projects