Recent Publications

. KAIST Multispectral Recognition Dataset in Day and Night. In T-ITS, 2018.


. StairNet: Top-Down Semantic Aggregation for Accurate One Shot Detection. In WACV, 2018.


. Multispectral Transfer Network: Unsupervised Depth Estimation for All-day Vision. In AAAI, 2018.

PDF Slides

. Thermal Image Enhancement using Convolution Neural Network. In IROS, 2016.

PDF Video

. Fast Multiple Objects Detection and Tracking Fusing Color Camera and 3D LIDAR for Intelligent Vehicles. In URAI, 2016.

PDF Project Video

. Thermal-Infrared based Drivable Region Detection. In IV, 2016.

PDF Project

. Geometrical Calibration of Multispectral Calibration. In URAI, 2015.


. Low-Cost Synchronization for Multispectral Cameras. In URAI, 2015.


. ARTRIEVAL: Painting Retrieval Without Expert Knowledge. In ICIP, 2015.

PDF Project

. All-Day Visual Place Recognition: Benchmark Dataset and Baselines. In CVPR-VPRICE, 2015.

PDF Project Poster Slides

. Multispectral Pedestrian Detection: Benchmark Dataset and Baselines. In CVPR, 2015.

PDF Project

. A Two Phase Approach for Pedestrian Detection. In ACCV-IVVT, 2014.


. Evaluation of Vocabulary Trees for Localization in Robot Applications. In ICCAS, 2013.



Multispectral Recognition Dataset

This dataset contains various illumination conditions (day, night, sunset, and sunrise) of multimodal data, which are of particular interest in autonomous driving-assistance tasks such as localization (place recognition, 6D SLAM), moving object detection (pedestrian or car) and scene understanding (drivable region).

Multispectral Pedestrian Benchmark

We developed imaging hardware consisting of a color camera, a thermal camera and a beam splitter to capture the aligned multispectral (RGB color + Thermal) images. With this hardware, we captured various regular traffic scenes at day and night time to consider changes in light conditions.


  • Depth Estimation Based on Thermal Image, and Neural Network Learning Method Korea, Pending (10-2017-0023475), Feb 2017.

  • Pattern Board for Geometrical Calibration in Multi-spectral Camera System Korea, Pending (10-2016-0122950), Sep 2016.

  • Thermal Image Enhancement using Deep Learning Algorithm Korea, Pending (10-2016-0100058), Aug 2016.

  • Detecting Smoke from Image US20150030203 (Filed on May 2014).


  • Honorable mention ($ 2,000 / Accept rate ~6.5%), The 24th HumanTech Paper Award, Samsung, Feb 2018

  • Gold prize ($ 10,000 / Accept rate 0.7%), The 23th HumanTech Paper Award, Samsung, Feb 2017

  • Bronze prize, Korea Invention Patent Exhibition (KINPEX), Dec 2016

  • 1st place, NVidia Deep Learning Contest, NVidia Korea, Nov 2016