About Me

Resume - JunHa Song


I’m Junha Song, a first-year MS.D student at KAIST, advised by Prof. In So Kweon and a research engineer at HYUNDAI MOBIS. Please visit my Git-blog for my research note from which you can see that I am a highly self- motivated student. My research interest lies in the intersection of machine learning and computer vision. I always dream that my research can help many people someday.

Juna Song


  • 2021 ~ Present
    KAIST - Division of Future Vehicle
    RCV Lab under Prof. In So Kweon
    M.S. Candidate
  • 2015 ~ 2020
    Kookmin University (Seoul, Korea)
    Department of Automobile and IT Convergence

Official Grade

  • University Major Credit
    4.43 / 4.5 (Total College Grade: 4.4/4.5)
    2018.08 - 880 / 990
    2020.08 - 825 / 990
  • OPIC
    2020.07 - IH (Intermediate High)
    2022.01 - 150 (Lv6)

Research Interests

  • Self-supervised learning such as CLIP and MAE
  • Test time adaptation for image segmentation
  • Reducing the efforts of the human annotator with domain adaptation or active learning But not limited to.

Awards and Honors

  • Vehicle monitoring over internet toward digital twins, Honorable Judge Award (Future Transport Design Award), Cloud Programming World Cup, Japan (2019)
  • Capstone Awards, The Korean Society of Automotive Engineers, Korea (2019)
  • National Science and Engineering Scholarship (Full Tuition), Korea Scholarship Foundation (2018 ~ 2020)


  • Detection of Surrounding Vehicles using Deep Neural Network and Fusion of Panoramic Camera and Lidar Sensor, KOFAC(한국과학창의재단), Korea (2019)
  • Satellite image precision object detection, DACON(국방과학연구소), Korea (2020)
  • Development of robust segmentation network having environment invariance, HYUNDAI MOBIS(현대모비스), Korea (2021)
  • Development of real-time masking/unmasking system for personal video information for public services such as CCTV (CCTV 등 공공서비스용 개인영상정보 실시간 마스킹/언마스킹 시스템 개발, News1, News2), Ministry of Science and ICT(과학기술정보통신부), Korea (2021~)

Research Experiences

  • Research Intern at Intelligence and Interaction Lab, Kookmin University (Feb. 2019 ~ Nov. 2019)
  • Research Intern at Machine Intelligence Lab, Kookmin University (Dec. 2019 ~ Present)
  • Junha Song, et al, “센서 퓨전과 통신을 이용한 차량의 디지털 트윈의 구현”, The Korean Sociey of Automotive Engineers, Korea (2019)
  • Myunghak Lee, Junha Song, “Style Transfer Maps from Satellite Images by using Generative Models”, KICS, Korea (2020)
  • (ECCV 2022 Submitted) J Song, C Zang, K Pack, I Shin, S Woo, I Kweon. “CD-TTA: Compound Domain Test Time Adaptation for Semantic Segmentation”


  • Programming language: Python, C++
  • Machine Learning Librarie: Pytorch, Tensorflow
  • Application development: ROS (Robot Operating System)
  • Sensor utilization: Camera, RGB-D Camera, LIDAR, GPS/IMU

© All rights reserved By Junha Song.