Shun-Hsiang Hsu

Shun-Hsiang Hsu / Publications

Publications

Research on synthetic data, vision-language learning, LLM evaluation, and visual inspection. Manuscripts under review are labeled.

2026

  1. Research illustration for Beyond Cracks: Synthetic Image and Geometry Generation for Computer Vision Detection and Severity Assessment of Diverse Concrete Surface Defects

    Beyond Cracks: Synthetic Image and Geometry Generation for Computer Vision Detection and Severity Assessment of Diverse Concrete Surface Defects

    Shun-Hsiang Hsu, Mani Golparvar-Fard

    Automation in Construction · 2026

    Under review

    Uses controllable 3D defect geometry to generate aligned imagery and supervision for recognizing concrete defects and assessing their severity. Manuscript under review.

    Contribution: First author; developed the geometry-grounded synthetic-data framework.

    Project

2025

  1. Research illustration for Self-Evaluation of Single and Multi-Agent LLMs as Assistants in Inspection Reporting Workflows

    Self-Evaluation of Single and Multi-Agent LLMs as Assistants in Inspection Reporting Workflows

    Shun-Hsiang Hsu, Yoonhwa Jung, Junryu Fu, Mani Golparvar-Fard

    Joint CSCE Construction Specialty & CRC Conference · 2025

    Equal contribution: Shun-Hsiang Hsu and Yoonhwa Jung

    Compares single- and multi-agent vision-language workflows for inspection reporting and studies model-generated self-assessment of report quality.

    Contribution: Co-first author; evaluated reporting workflows and self-assessment strategies.

    Paper (PDF)

2024

  1. Research illustration for VisualSiteDiary: A detector-free Vision-Language Transformer model for captioning photologs for daily construction reporting and image retrievals

    VisualSiteDiary: A detector-free Vision-Language Transformer model for captioning photologs for daily construction reporting and image retrievals

    Yoonhwa Jung, Ikhyun Cho, Shun-Hsiang Hsu, Mani Golparvar-Fard

    Automation in Construction · 2024

    Equal contribution: Yoonhwa Jung, Ikhyun Cho, and Shun-Hsiang Hsu

    A detector-free vision-language Transformer generates construction photolog captions and supports image retrieval from visual observations.

    Contribution: Equal contributor with Yoonhwa Jung and Ikhyun Cho; co-developed VisualSiteDiary.

    Paper
  2. Research illustration for VL-Con: Vision-Language Dataset for Deep Learning-based Construction Monitoring Applications

    VL-Con: Vision-Language Dataset for Deep Learning-based Construction Monitoring Applications

    Shun-Hsiang Hsu, Junryu Fu, Mani Golparvar-Fard

    ISARC. Proceedings of the International Symposium on Automation and Robotics in Construction · 2024

    Builds construction-specific image-text pairs to study the domain gap in vision-language models and support construction-monitoring applications.

    Contribution: First author; developed VL-Con and its domain-specific vision-language data approach.

    Paper Paper (PDF) Code
  3. Research illustration for Requirements for Parametric Design of Physics-Based Synthetic Data Generation for Learning and Inference of Defect Conditions

    Requirements for Parametric Design of Physics-Based Synthetic Data Generation for Learning and Inference of Defect Conditions

    Shun-Hsiang Hsu, Mani Golparvar-Fard

    Construction Research Congress 2024 · 2024

    Examines parametric modeling and rendering requirements for synthetic defect data, with a crack-segmentation study evaluating transfer to real images.

    Contribution: First author; research on parameterized defect geometry and synthetic-data generation.

    Paper Read more

2023

  1. Research illustration for Defect inspection of indoor components in buildings using deep learning object detection and augmented reality

    Defect inspection of indoor components in buildings using deep learning object detection and augmented reality

    Shun-Hsiang Hsu, Ho-Tin Hung, Yu-Qi Lin, Chia-Ming Chang

    Earthquake Engineering and Engineering Vibration · 2023

    Combines deep-learning defect detection with augmented reality to locate and measure indoor building defects in a 3D inspection environment.

    Contribution: First author.

    Paper

2022

  1. Research illustration for Impacts of label quality on performance of steel fatigue crack recognition using deep learning-based image segmentation

    Impacts of label quality on performance of steel fatigue crack recognition using deep learning-based image segmentation

    Shun-Hsiang Hsu, Ting-Wei Chang, Chia-Ming Chang

    Smart Structures and Systems, An International Journal · 2022

    Studies how image cropping and annotation refinement affect steel fatigue crack segmentation, providing evidence for dataset-preparation choices.

    Contribution: First author.

    Paper Read more

2021

  1. Research illustration for Concrete surface crack segmentation based on deep learning

    Concrete surface crack segmentation based on deep learning

    Shun-Hsiang Hsu, Ting-Wei Chang, Chia-Ming Chang

    European Workshop on Structural Health Monitoring · 2021

    Compares DeepLabv3+ and Mask R-CNN for concrete crack segmentation and derives geometric measurements from the predicted masks.

    Contribution: First author.

    Paper