Automation in Construction
Seeing the smallest safety equipment
Tiny hooks and straps are easily missed in construction surveillance images.
DSDL improves training assignments and bounding-box regression for tiny objects.
CIVIL & ENVIRONMENTAL ENGINEERING
SEOUL, SOUTH KOREAIntelligence for
the infrastructure
we depend on.
Computer vision, physics-informed AI, and engineering agents for construction safety and infrastructure inspection.
Explore our research Join Us ↗THE RESEARCH APPROACH01—03
01
Observe
Images · 3D geometry · Sensor data
02
Understand
Computer vision · Physics · AI
03
Act
Safety · Inspection · Maintenance
From real-world observations to informed engineering decisions.
Our Vision
Dream Big.
Aim for the Best.
Build a Better World.
We aspire to be a world-leading research group where ambitious people pursue bold ideas, grow into exceptional researchers and leaders, and shape the future of construction and infrastructure for a better world.
큰 꿈을 품고 세계 최고에 도전하며, 탁월한 연구자와 리더를 길러내어 건설과 인프라의 미래를 통해 더 나은 세상을 만든다.
Automation in Construction
Retrieval that understands construction documents
General-purpose retrieval can miss the meaning of specialized Korean construction documents.
Contrastive sentence generation and Matryoshka representations improve domain-specific retrieval.
Automation in Construction
Keeping thin cracks connected
Thin cracks are easily lost or disconnected in segmentation predictions.
FACS-Net and crack topology loss address frequency information and crack connectivity.
Automation in Construction
Understanding the context of construction safety
Safety assessment depends on the relationships between workers, equipment, and surroundings.
Large vision-language models are optimized for context-aware construction safety assessment.
01 / RESEARCH
Built on engineering.
Advanced through AI.
Methods that connect visual observations, physical processes, and engineering knowledge.
Visual intelligence
for infrastructure
Computer vision for damage assessment, crack segmentation and matting, and automated inspection of civil infrastructure.
3D sensing &
reconstruction
Photogrammetry, image enhancement, and 3D modeling across terrestrial and underwater environments.
Physics-informed
intelligence
Integrating physical principles with machine learning to reconstruct pore water pressure fields and support embankment failure detection.
AI for engineering
decisions
Multimodal models, retrieval-augmented generation, and AI agents for construction safety, engineering documents, and design calculations.
Fall 2026 undergraduate teaching
Undergraduate courses: Artificial Intelligence for Civil and Environmental Engineering, and Engineering Information Processing.
Read more →Journal Paper Publication
Seokhwan Kim, Taegeon Kim, Kichang Choi, Siheon Joo, Hongjo Kim's paper 'Tiny object detection using Distance-guided, Signed, and Densified Learning (DSDL) for construction site safety monitoring' has been published in 'Automation in Construction'.
Promotion
Prof. Hongjo Kim has been promoted to Associate Professor at the Department of Civil and Environmental Engineering, Yonsei University.
Journal Paper Publication
Minkyu Koo, Taegeon Kim, Minhyun Lee, Kinam Kim, Hongjo Kim's paper 'Domain-adaptive instance segmentation for far-field object monitoring using SAM-based weak supervision and noisy student self-training' has been published in 'Automation in Construction'.
Research and Academic Writing for Graduate Students, Part 1: What Is Research?
When does building an LLM system or improving crack segmentation become research? Start with the question, the evidence, and the contribution.
Read more ↗
Can AI Check Structural Calculations? How RAMASC Retrieves, Calculates, and Reviews
Beyond a convincing answer: connecting design references, Python execution, and multiple AI reviewers for structural checks.
Read more ↗
How Much Should a Site Invest in Safety Technology? Exploring Break-Even with ECOSTAT
Beyond detection accuracy: how project scale, implementation costs, and accident-prevention effectiveness shape an investment decision.
Read more ↗
A shared curiosity. A collective effort. ↗The People We Seek
Aim High. Grow Together. Make It Matter.
Aim High.
People who refuse to settle, strive to become better versions of themselves, and continually challenge themselves to pursue big dreams and ambitious goals.
Grow Together.
People who contribute to the growth of those around them as well as their own, helping one another reach greater heights together.
Make It Matter.
People who look beyond personal success, use their knowledge and abilities to serve people and society, and seek to contribute to the common good and a better world.
05 / CONTACT
Let’s connect
ideas to infrastructure.
For research collaboration and graduate research inquiries, contact Professor Hongjo Kim.
hongjo@yonsei.ac.kr Department of Civil & Environmental EngineeringYonsei University
Seoul, South Korea