Home

Awesome

Edge-guided Multi-domain RGB-to-TIR image Translation for Training Vision Tasks with Challenging Labels

Accepted Proceedings to ICRA 2023

<div align="left"> <a href="https://scholar.google.com/citations?user=u6VDnlgAAAAJ&hl=ko&oi=ao">Dong-Guw Lee</a>, <a href="https://scholar.google.co.kr/citations?user=ivOqySYAAAAJ">Myung-Hwan Jeon</a>, <a href="https://scholar.google.com/citations?user=W5MOKWIAAAAJ&hl=ko&oi=ao">Younggun Cho</a>, <a href="https://ayoungk.github.io/">Ayoung Kim</a> at <a href="https://rpm.snu.ac.kr">RPM Robotics Lab</a> </div>

Overview of the edge-guided multi-domain RGB2TIR translation network

<div align="center">

overview_new-1

</div>

Proposed pipeline for training vision tasks with challenging labels

<div align="center">

proposed_method-1

</div>

Results

Disclaimer

-The same model was used for both synthetic and real RGB to TIR image translation

-The model was trained on identical datasets (sRGB=GTA, TIR=STheReO)

Results on synthetic RGB to TIR translation

<div align="center">

synthetic_rgb_original-1

</div>

Results on real RGB to TIR translation

<div align="center">

real_rgb_translation_pdf-1

</div>

Results on thermal optical flow estimation using the proposed method

<div align="center">

optical_flow_comparison-1

</div>

Video demonstration

Video Label

https://youtu.be/zq8Qh9ygm6w

TODO

Environment Setup

How To Use: RGB to TIR translation

Please download them from here: {link to google drive}

*If the link doesn't work, please file an issue!

Network Details

Edge-guided multi-domain RGB2TIR translation architecture

Citation

Please consider citing the paper as:

@ARTICLE{lee-2023-edgemultiRGB2TIR,
author={Lee, Dong-Guw and Kim, Ayoung},
conference={IEEE International Conference on Robotics and Automation}, 
title={Edge-guided Multi-domain RGB-to-TIR image Translation for Training Vision Tasks with Challenging Labels}, 
year={2023},
status={underreview}

Also, a lot of the code has been built on top of MUNIT (ECCV2018), so please go cite their paper as well.

Contact

If you have any questions, contact here please

donkeymouse@snu.ac.kr