Geonho Bang

Autonomous Driving Researcher (PhD Candidate)

About Me

Hi, my name’s Geonho Bang and I’m a PhD student at the SPA Lab under Professor Junwon Choi at Seoul National University. My research focuses on perception systems for autonomous driving, specifically 3D object recognition using radar sensors. Throughout my PhD journey, I have worked on advanced radar-based perception algorithms to enhance the safety and efficiency of self-driving cars. I have had the opportunity to collaborate with leading automotive companies such as Hyundai Motors and HL Klemove, which has provided me with valuable industry insights and practical experience. My work has been recognized at prestigious conferences like CVPR and ICRA, where I have submitted several research papers.

I am most skilled in: Autonomous Driving, Knowledge Distillation, Radar-based 3D Object Detection, and Machine Learning

Overview

Research interest areas:
  • Perception for intelligent vehicles and mobile robots
  • Knowledge Distillation for 3D object detection
  • Radar, Camera and LiDAR-based 3D object detection
  • Sensor fusion-based 3D object detection and tracking
  • Radar based point cloud generation

Education

Ph.D. in Artificial Intelligence (Advisor: Prof. Jun Won Choi)

Hanyang University, Seoul, South Korea

Mar 2024 - Present


M.S. in Artificial Intelligence (Advisor: Prof. Jun Won Choi)

Hanyang University, Seoul, South Korea

Mar 2022 - Feb 2024


B.S. in Automobile and IT Convergence

Kookmin University, Seoul, South Korea

Mar 2016 - Feb 2022

Publications

RadarDistill: Boosting Radar-based Object Detection Performance via Knowledge Distillation from LiDAR Features

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

Geonho Bang, Kwangjin Choi*, Jisong Kim, Minjae Seong, Dongsuk Kum and Jun Won Choi


PillarGen: Enhancing Radar Point Cloud Density and Quality via Pillar-based Point Generation Network

IEEE International Conference on Robotics and Automation (ICRA), 2024.

Geonho Bang, Jisong Kim*, Kwangjin Choi, Minjae Seong, Jaechang Yoo, Eunjon Pyo and Jun Won Choi


RCM-Fusion: Radar-Camera Multi-Level Fusion for 3D Object Detection

IEEE International Conference on Robotics and Automation (ICRA), 2024.

Jisong Kim, Minjae Seong, Geonho Bang, Dongsuk Kum, and Jun Won Choi

(* indicates equal contributions)

Projects

Development of 3D Object Detection Network using Camera-Radar-LiDAR Sensor Fusion

Hyundai Motor Group (Apr 2024 - Apr 2025)

  • Develop a Lightweight Radar-Camera Sensor Fusion Model
  • Develop Knowledge Distillation Methods for Radar-Camera Sensor Fusion Model


Development of the basic technology for AR smart glasses that can be used for prehospital advanced life support

Hanyang University Guri Hospital (Oct 2023 - Feb 2024)

  • Collect the Cardio Pulmonary Resuscitation (CPR) compression depth measurement and image using Ambu manikin and smartphone
  • Design the deep learning based Cardio Pulmonary Resuscitation (CPR) compression depth estimation network using smartphone image


Radar Point Cloud Perception Network Development

HL Klemove (Jul 2022 - Jun 2023)

  • Develop the radar point cloud generation network to enhance point density and quality
  • Design the radar-based 3D object detector using knowledge distillation from high-quality radar to low-quality radar


Development of Distributed On-Chip Memory-Compute Integration PIM Semiconductor Technology for Edge Use

DEEPX (Apr 2022 - Dec 2022)

  • Deploy the 3D object detection model to NVIDIA Jetson AGX Xavier
  • Optimize the 3D object detection model using the TensorRT library

Review Experiences

  • Conference Review : ICRA
  • Journal Review : IEEE TVT

Patents

PILLAR BASED POINT CLOUD GENERATION METHOD AND APPARATUS THEREOF (KR - Application No.10-2024-0033128)

Jun Won Choi, Geonho Bang, Jisong Kim


3D OBJECT DETECTION APPARATUS AND METHOD THEREOF (KR - Application No.10-2024-0033147)

Jun Won Choi, Geonho Bang, Jisong Kim, Kwangjin Choi

Skills

  • Computer Languages : Python, C++
  • Deep Learning Tools : Pytorch, Tensorflow
  • Language : Korean (Native), English (Intermediate)