Gihwan Kim (김기환)

Research Interest

Research Areas: Model Compression, Model Acceleration and TinyML

Research Topics:

• • Model Acceleration and Compression

  • Computer Vision

  • Edge computing

  • On-device AI for drones & Satellites

Education

• 2025 - Present: Visiting Scholar, Carnegie Mellon University, Software and Societal Systems Department, USA

• 2024 - 2026 M.S Dept. of Architectural Engineering, Chungnam National University (Advisor: Hyungshin Kim)

• 2017 - 2024 B.S Dept. of Architectural Engineering, Chungnam National University

Experience

• Jan 2024 - Feb 2024: Research Intern 

  • Electronics and Telecommunications Research Institute, Republic of Korea, Daejeon

•  Jan 2023 - Feb 2023: Visiting Scholar

  • Purdue University, USA

  • Participated as a Korean Software Square Winter 2023 program.

  • Implemented and researched counter drone system with on-device AI.

• Apr 2022 - Dec 2023: Research Student

  • Embedded Systems Laboratory at Chungnam National University, Republic of Korea, Daejeon

  • Researched quantization and optimization for a low power computer vision system.

• Feb 2020 - Jun 2021: SW program 42 Seoul 

  • Innovation Academy, Republic of Korea, Seoul

  • Participated software education program.

Publication (CV)

• Kim, Gihwan, Jemin Lee, and Hyungshin Kim. "IPTQ-ViT: Post-Training Quantization of Non-linear Functions for Integer-only Vision Transformers." WACV 2026.

• Kim, Gihwan, et al. "Mixed Non-linear Quantization for Vision Transformers." European Conference on Computer Vision. Cham: Springer Nature Switzerland, 2024.

  • Model compression and quantization

• Lee, Hanbin, et al. "Lightweight Disaster Semantic Segmentation for UAV On-Device Intelligence." IGARSS 2024, IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2024 (2nd author).

  • Model compression and quantization

• Kim, Gihwan, et al. "Power Efficient Long Range Drone Networking System for UAV Detection." 2023 14th International Conference on Information and Communication Technology Convergence (ICTC). IEEE, 2023.

  • Drone detection system

• Kim, Gihwan, et al. (2023). "Lightweight Semantic Segmentation Model with ResNet-based Encoder-Decoder Architecture for Edge Device"

  • Model compression and quantization

Skills

• Deep learning & Artificial Intelligence

  • Python

  • PyTorch, CUDA, TensorRT

  • Semantic Segmentation, Object Detection, Image Classification

• Robotics

  • C/C++

  • FreeRTOS

  • Computer Vision

Outside of Work

• Beyond my academic interests, I find joy in swimming and traveling !! They refresh my mind and broaden my perspective.

• 🌍 Places I’ve Visited 

  • 🇮🇹 Italy: Milan, Genoa, Dolomites 

  • 🇫🇷 France: Paris 

  • 🇬🇷 Greece: Athens, Santorini 

  • 🇺🇸 United States: West Lafayette, San Francisco, Santa Clara, Chicago

  • 🇯🇵 Japan

• 🌍 Places I planned 

  •