About

Dr. Mani Golparvar-Fard is a Professor of Civil Engineering, Computer Science, and Technology Entrepreneurship at the University of Illinois at Urbana-Champaign (UIUC). He is also a Faculty Entrepreneurial Fellow, an Excellence Faculty Fellow, and the director of the Real-time and Automated Monitoring and Control (RAAMAC) lab. His educational background includes a Bachelor of Science in Civil Engineering and a Master of Science in Civil Engineering from Iran University of Science and Technology, a Master of Applied Science in Civil Engineering from the University of British Columbia, and both a Master of Computer Science and a Ph.D. in Civil Engineering from UIUC. Prior to his current position, he served as an Assistant Professor at Virginia Tech. Dr. Golparvar-Fard has extensive experience working with national and international construction companies, notably with Turner Construction. His research focuses on integrating civil engineering, computer science, and data science to develop innovative solutions for construction monitoring, automation, and data analytics. He has several patents and is involved with Reconstruct Inc., a startup company based on his research, which has raised significant venture funding and has been recognized for its innovative solutions in construction technology. His contributions have been recognized through numerous awards, including the 2018 Walter Huber Research Prize from the American Society of Civil Engineers and the 2017 Young Professional Award from Engineering News-Record. Dr. Golparvar-Fard serves on the editorial boards of several prominent journals and is actively engaged in advancing research and innovation in the construction and civil engineering fields.

Selected publications

• Voice-Assisted AI Chatbot for Construction: Speech-to-Text Deep Learning Transformer Trained with Synthetic Data Sets

  2025-12-11 · 1 citations

  articleSenior authorCorresponding

  Effectively managing information workflows, encompassing design, construction, and resource utilization data, is crucial for operational success of construction projects. To enable real-time and seamless access to such information on and offsite, this paper presents a Voice-activated Artificial Intelligence (AI) assistant, ConstructVoiceBot. Our solution leverages the first domain-specific Speech-to-Text Transformer model, called Con-Whisper, fostering seamless interaction between construction domain users and an AI agent. This voice-activated AI agent—trained and validated with a synthetic voice-text data set from 35 commercial building projects—harnesses domain-specific knowledge, generating precise voice-based prompts for a generative language model within the chatbot. To further improve performance, ConstructVoiceBot is enhanced with a document parser and a pre-trained vision-language model. Experimental results highlight the data set’s efficacy in enhancing speech-to-text model performance compared to general commercial models. We delve into practical applications through use cases such as Time and Material reporting, daily construction reporting, quality assurance, and curation of construction workflows.

  PublisherDOI

• ExpertScheduler: A Mixture-of-Experts Transformer Language Model for Generating Tasks in a Construction Look-ahead Plan

  SSRN Electronic Journal · 2025-01-01

  preprintOpen accessSenior author
  PublisherDOI

• Hybrid deep learning model for predicting failure properties of asphalt binder from fracture surface images

  Computer-Aided Civil and Infrastructure Engineering · 2025-02-20 · 4 citations

  articleOpen access

  Cracking impacts asphalt concrete durability primarily due to cohesive asphalt binder failures. The poker chip test has recently been introduced to better characterize the cracking potential of asphalt binders by fracturing a specimen in a realistic stress state to a thin binder film. However, broader adoption faces challenges due to high instrumentation costs for measuring load and displacement. This paper presents and validates a deep learning model that predicts ductility and tensile strength from posttest images of fractured binder surfaces, with potential extensions to simplified instrumentation. The hybrid model, named PCNet, integrates a custom lightweight convolutional neural network (CNN) developed to capture local features (e.g., edges, boundaries, contours) within fracture cavities with a Swin Transformer that models global contextual dependencies. A bidirectional cross-attention fusion module is designed to facilitate mutual information exchange between CNN and transformer branches. The fused features are then processed by a fully connected network (FCN) to predict indices derived from the test. The proposed model demonstrates high predictive accuracy across a range of binders and test configurations, achieving an R 2 $ R^2$ of 0.966 and a mean absolute percentage error (MAPE) of 12.95% in predicting ductility, while also attaining an R 2 $ R^2$ of 0.947 and a MAPE of 9.15% for strength, outperforming standalone models. Monte Carlo Dropout is also incorporated in the FCN to quantify prediction confidence. This cost-effective methodology provides insights into fracture propagation in soft viscoelastic media and contributes to the field of experimental mechanics. With further data collection, the model holds potential for broader implementation, directly linking fracture surface images and mixture or field-scale cracking behavior.

  PublisherOA PDFDOI

• The True Environmental Cost of Data Centers: An Insight into the Construction Phase of the Era of Artificial Intelligence

  CIB Conferences · 2025-06-19

  articleOpen accessSenior author

  In recent years, there has been a significant adoption of commercial applications of Artificial Intelligence (AI). These applications have created a demand for infrastructure that support data storage, mining, and throughput compute power services, commonly known as data centers. This demand, which is projected to substantially increase over the next decade, places data centers to be a major contributor to the construction industry's global carbon emissions. However, with research mostly focusing on minimizing carbon emissions during the operation phase of most buildings, environmental impacts incurred during their construction phase are still deserving a deeper study. To address this gap, this work empirically evaluates the true cost to the environment of the construction of a data center by collecting and analyzing real-world project data on carbon emissions of material utilization and operation sequences. Findings of the presented analyses expose how on average carbon emissions are higher by 414% in the case of foundation systems (A - Substructure) for all data centers when compared to residential projects, and how the expected 25% increase of data center construction can pose an additional 1,832.51M kgCO2e in 2025, translating to 0.0054% of the world’s total carbon emissions coming from the U.S. market alone. This work aims to address sustainable development goals through understanding the extent of the environmental impact of building such critical infrastructure, which is vital in supporting the connectivity of people, fostering innovation, and accelerating industrialization.

  PublisherOA PDFDOI

• New Metrics to Benchmark and Improve BIM Visibility Within a Synthetic Image Generation Process for Computer Vision Progress Tracking

  Lecture notes in civil engineering · 2024-09-18 · 3 citations

  book-chapterSenior author
  PublisherDOI

• Evaluation of Mapping Computer Vision Segmentation from Reality Capture to Schedule Activities for Construction Monitoring in the Absence of Detailed BIM

  Proceedings of the ... ISARC · 2024-05-27 · 1 citations

  articleOpen accessSenior author

  Over the past few years, research has focused on leveraging computer vision in construction progress monitoring, particularly in comparing construction photologs to Building Information Modeling (BIM), with or without schedule data.The practical application of these techniques and a large number of startups that have brought hyper AI and human-in-the-loop services around progress monitoring have revealed several gaps: 1) Current BIM-driven projects do not have model disciplines at the right level of maturity and Level of Development; 2) definitions of states of work-in-progress that are detectable from images are not formalized; 3) poor schedule quality and lack of frequent progress update challenges the incorporation of detailed 4D BIM for progress tracking.Such gaps are addressed in this work by exploring the requirements for mapping modern computer vision techniques for object segmentation with construction schedule activities to automate progress monitoring applications using computer vision without BIM as a baseline.The approach utilizes reality mapping practices to offer time machines for construction progress, organizing photologs over space and time.Additionally, this work shows how Large Language Models can structure schedule activity descriptions around <Uniformat Object Classification, Location>, focusing on how vision and language models can be trained separately with limited annotated data.ASTM Uniformat classification is utilized to map triangulated object segments from images to color-coded 3D point clouds aligned with schedule activities without the need for image and language feature alignments.Exemplary results on tied new transformer-based models with few-shot learning are shown, and the requirements for full-scale implementation are discussed.

  PublisherOA PDFDOI

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

  Automation in Construction · 2024-05-30 · 30 citations

  articleOpen accessSenior author

  This paper presents VisualSiteDiary, a Vision Transformer-based image captioning model which creates human-readable captions for daily progress and work activity log, and enhances image retrieval tasks. As a model for deciphering construction photologs, VisualSiteDiary incorporates pseudo-region features, utilizes high-level knowledge in pretraining, and fine-tunes for diverse captioning styles. To validate VisualSiteDiary, a new image captioning dataset, VSD, is presented. This dataset includes many realistic yet challenging cases commonly observed in commercial building projects. Experimental results using five different metrics demonstrate that VisualSiteDiary provides superior-quality captions compared to the state-of-the-art image captioning models. Excluding the task of object recognition, the presented model also outperformed mPLUG –the state-of-the-art visual-language model– in the image retrieval task by 0.6% in precision and 0.9% in recall, respectively. Detailed discussions illustrate practical examples on how VisualSiteDiary improves the process of creating daily construction reports, paving the way for future developments in the field.

  PublisherDOI

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

  Proceedings of the ... ISARC · 2024-05-27 · 2 citations

  articleOpen accessSenior author

  VL-Con: Vision-Language Dataset for Deep Learning-based Construction Monitoring Applications Shun-Hsiang Hsu, Junryu Fu, Mani Golparvar-Fard Pages 1128-1135 (2024 Proceedings of the 41st ISARC, Lille, France, ISBN 978-0-6458322-1-1, ISSN 2413-5844) Abstract: Recently, vision-language research has gained significant interest by successfully connecting visual concepts to natural language, advancing computer vision-based construction monitoring using a wide variety of text queries. While vision language models demonstrate high capability, performance degradation can be expected when adapting the model to the ever-changing construction scenarios. In contrast to the source image-text pairs, it is more challenging to cover the multitude of potentially involved objects and their naming conventions for construction activities. To bridge the domain gap, this study aims to collect construction-specific image-text pairs of building elements and related site work based on the ASTM Uniformat II. The image-text pairs of 641 activities in Uniformat are retrieved from the LAION-5B dataset based on the image and text embeddings using CLIP with two different prompts. Then, the collected images are labeled at the image level to conclude the requirements of the vision-language datasets for further development. Based on the results, a vision-language dataset, VL-Con, consisting of image-text pairs for construction monitoring applications is proposed with the aid of a construction semantic predictor and prompt engineering. The proposed VL-Con dataset can be accessed through https://github.com/huhuman/VL-Con. Keywords: Vision-Language Dataset, Construction Monitoring, Foundation Model DOI: https://doi.org/10.22260/ISARC2024/0146 Download fulltext Download BibTex Download Endnote (RIS) TeX Import to Mendeley

  PublisherOA PDFDOI

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

  2024-03-18

  articleSenior authorCorresponding

  The advancement of Artificial Intelligence (AI)-driven defect detection has already demonstrated promises to boost quality assurance and control, as well as condition assessment in the built environment. However, training defect detection models requires hefty amounts of reality capture data, and labeling is considered expensive. In most cases, such data may not cover all situations of defects. Synthetic data, most recently made with Building Information Models (BIM), is turbocharging model development for learning defect features. Nevertheless, few studies focused on characterizing defects to classify their severity, which is crucial to the condition assessment. To that end, this study explores the requirements for generating synthetic data. Parametric physics-based modeling approaches are carefully examined. Using the underlying geometric properties of such data, the condition of each defect can be determined. The feasibility of synthetic defect data is validated with a case study of crack segmentation using the transformer-based model, SegFormer. Examples of how different scenarios can be generated photo-realistically with the use of physics-based rendering for creating varying geometrical characteristics, appearance, and viewpoints of defects are presented. The generated synthetic crack datasets can successfully be used to train the SegFormer model and reach promising predictions on real crack images.

  PublisherDOI

• Bi-Directional Image-to-Text Mapping for NLP-Based Schedule Generation and Computer Vision Progress Monitoring

  2024-03-18 · 2 citations

  articleSenior authorCorresponding

  State-of-the-art in construction document analytics and progress detection has experienced accelerated growth over the last decade. However, each area encountered isolated growth, not considering their interactions. Today, progress monitoring practices are often neglected due to requiring manual input of visible progress against schedules. Such a challenge can be attributed to (1) vision-based progress tracking lacking formal construction work templates applied in common construction workflows, and (2) research in automated schedule generation and analytics lacking focus on extracting fragnets from a body of existing schedules. This study brings together insights on research trends for automated schedule generation and analytics using Natural Language Processing (NLP) and detection of under-construction objects using Computer Vision. Finally, the AIConstruct system is presented to demonstrate, for the first time, how the integration of text and image can create seamless data synchronization for construction progress monitoring and automated schedule generation, unlocking a new research paradigm.

  PublisherDOI

Labs

• Real-time and Automated Monitoring and Control (RAAMAC) labPI

Awards & honors

• 2018 Walter Huber Research Prize from the American Society o…
• 2017 Young Professional Award from Engineering News-Record (…
• 2016 ASCE Dan H. Halpin Award for Scholarship in Constructio…
• 2013 ASCE James R. Croes Medal for innovation in Civil Engrg
• 2013 FIATECH CETI award for outstanding researcher