Sangmin Hong

I am a PhD student at the Computer Vision Lab, part of the IPAI at Seoul National University, where I work on 3D computer vision. My PhD advisor is Kyoung Mu Lee.

Before emarking on my doctoral studeis, I completed my Bachelor of Science in Electronic Engineering at the University of Manchester, where I developed a keen interest in the control theory and robotic systems. Following my undergraduate degree, I further specialized by pursuing a Master of Science in Advanced Control and System Engineering from 2019 to 2020. My master's research centered on developing efficeint alogrithms for path planning in autonomous vehciles, aiming to enhance navigational technologies in self-driving cars.

My academic journey reflects a deep commitment to understanding and innovating within the realm of intelligent systems, when I continually strive to blend theoretical research with practical applciations in artificial intelligence and robotics.

Email  /  GitHub  /  Google Scholar  /  LinkedIn

Research

I'm interested in computer vision, 3d Vision, and AI for Manufacturing

Geo2Gcode: Bridging 3D Geometry and Machining via Direct Manufacturing Toolpath Generation

Geo2Gcode generates manufacturing toolpaths directly from 3D geometry, closing the gap between a geometric model and the machine instructions that produce it.

Test-Time Part Generation and Assembly Optimization with Hierarchical Guidance from Large Language Models

We generate object parts and optimize their assembly at test time, using hierarchical guidance from large language models to decide what parts a target object needs and how they fit together.

Text2Tech: Text to 3D CAD Generation via Technical Drawings

Text2Tech translates a textual description into isometric renderings, converts them into orthographic technical drawings, and reconstructs a functional 3D CAD model from those views.

PrintAnything: Learning Geometric Plan Map for 3D Printing G-code Generation from Unoriented Point Clouds

PrintAnything turns a raw, unoriented point cloud straight into executable 3D printing G-code without any mesh reconstruction, by predicting a geometric plan (G-plan) map of per-slice occupancy, region and flow.

StarryGazer: Leveraging Monocular Depth Estimation Models for Domain-Agnostic Single Depth Image Completion

StarryGazer completes a sparse depth image without ground-truth depth, using a pre-trained monocular depth estimation model to synthesize training pairs so that the refinement network generalizes across domains.

CNC-Net: Self-Supervised Learning for CNC Machining Operations

We introduce a pioneering approach named CNC-Net, representing the use of deep neural networks (DNNs) to simulate CNC machines and grasp intricate operations when supplied with raw materials.

ACL-SPC: Adaptive Closed-Loop system for Self-Supervised Point Cloud Completion

ACL-SPC takes a single partial input and attempts to output the complete point cloud using an adaptive closed-loop (ACL) system that enforces the output same for the variation of an input.

Awards & Scholarships

2026	1st Place, Zero-shot Accident Anticipation Competition, AUTOPILOT Workshop, CVPR 2026 SNU CVLab Team — Sangmin Hong, Kyoung Mu Lee
2026	Outstanding Paper Award (Bronze Prize), 38th Workshop on Image Processing and Image Understanding (IPIU)
2026	Finalist, KRAFTON AI R&D Hackathon
2024	IPAI Scholarship, Seoul National University
2021–2023	AI Graduate School Program Fellowship, IITP, Government of Korea