Seunghyun Kim

My background is in Electronics, Information & Communications Engineering, with a focus on computer vision and real-time AI systems.

My experience includes deep learning-based perception, model inference, and system integration under real-world constraints, particularly where accuracy and latency need to be considered together.

During my undergraduate research, I developed a YOLOv8 + MLP-based accident risk classification system using lane-specific vehicle speed estimation, achieving 89% accuracy and 5 ms inference. The work was published as a first-author journal paper.

At ETRI, I worked on a 5G-based EO/IR drone object detection system, covering image preprocessing, model inference, optimization, and system validation. The system achieved 0.6 s end-to-end latency and improved detection accuracy through EO/IR fusion.

These experiences shaped my interest in building practical AI systems that connect data processing, models, and software for real-world applications.

I am particularly interested in AI software development for autonomous systems, drones, and intelligent applications.