Colorization of Depth Map via Disentanglement (ECCV 2020 accepted)

This is the PyTorch implementation for our ECCV'20 paper:

Colorization of Depth Map via Disentanglement PAPER

Chung-Sheng Lai, Zunzhi You, Ching-Chun Huang, Yi-Hsuan Tsai, Wei-Chen Chiu

Environment

1. Python 3.7.1
2. PyTorch 1.0.0
3. CUDA 9.1
4. Ubuntu 18.04.2 LTS

Installation

git clone https://github.com/alanlai199/ColorizeDepthNet.git
cd ColorizeDepthNet

download dataset in the folloing link link and unzip the file under folder named data

Training

train with random flipping and TIP

python train.py

train with Full Model setting

python train_full.py

Colorized image

colorized all depth maps in testing set with abitary reference RGBD pairs in data/nyu2_ref.csv

argument num_of_ref in generate.py needs to be changed

python generate.py

Citation

@InProceedings{lai_2020_ECCV,
author = {Lai, Chung-Sheng and You, Zun-Zhi and Huang, Ching-Chun and Tsai, Yi-Hsuan and Chiu, Wei-Chen},
title = {Colorization of Depth Map via Disentanglement},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
month = {August},
year = {2020}
}