Article

A critical review of vision-based occupational health and safety monitoring of construction site workers

June 2020
Safety Science 126(7):104658

DOI:10.1016/j.ssci.2020.104658

Authors:

Rui Shi

Dalian University of Technology

Zhen Yang

Dalian University of Technology

Globally, the occupational health and safety (OHS) of construction workers has long been a serious concern. To address this issue, there is an urgent need for an efficient means to continuously monitor the construction site to eliminate potential hazards in a timely manner. As robust and automated video and image information extraction and processing tools for construction sites, computer vision techniques have been considered to be effective solutions and been applied for the occupational health and safety monitoring of construction site workers. This paper aims to use bibliometric and content-based analysis methods to review the previous attempts in related fields and present the current research status in this field. The results clarify the major limitations and challenges of the current research from both technical and practical perspectives, in turn suggesting the direction of future research.

DTR: A Unified Detection-Tracking-Re-identification Frame- work for Dynamic Worker Monitoring in Construction Sites

Conference Paper

Jul 2024

The detection and tracking of construction workers in building sites generate valuable data on unsafe behavior, work productivity, and construction progress. Many computer vision-based tracking approaches have been investigated and their capabilities for tracking construction workers have been tested. However, the dynamic nature of real-world construction environments, where workers wear similar outfits and move around in often cluttered and occluded regions, has severely limited the accuracy of these methods. Herein, to enhance the performance of vision-based tracking, a new framework is proposed which seamlessly integrates three computer vision components: detection, tracking, and re-identification (DTR). In DTR, a tracking algorithm continuously tracks identified workers using a detector and tracker in combination. Then, a re-identification model extracts visual features and utilizes them as appearance descriptors in subsequent frames during tracking. Empirical results demonstrate that the proposed method has excellent multi-object-tracking accuracy with better accuracy than an existing approach. The DTR framework can efficiently and accurately monitor workers, ensuring safer and more productive dynamic work environments.

Innovations in safety management for construction sites: the role of deep learning and computer vision techniques

Article

Full-text available

Nov 2024
Construct Innovat

Purpose-The purpose of this study is to investigate the potential of using computer vision and deep learning (DL) techniques for improving safety on construction sites. It provides an overview of the current state of research in the field of construction site safety (CSS) management using these technologies. Specifically, the study focuses on identifying hazards and monitoring the usage of personal protective equipment (PPE) on construction sites. The findings highlight the potential of computer vision and DL to enhance safety management in the construction industry. Design/methodology/approach-The study involves a scientometric analysis of the current direction for using computer vision and DL for CSS management. The analysis reviews relevant studies, their methods, results and limitations, providing insights into the state of research in this area. Findings-The study finds that computer vision and DL techniques can be effective for enhancing safety management in the construction industry. The potential of these technologies is specifically highlighted for identifying hazards and monitoring PPE usage on construction sites. The findings suggest that the use of these technologies can significantly reduce accidents and injuries on construction sites. Originality/value-This study provides valuable insights into the potential of computer vision and DL techniques for improving safety management in the construction industry. The findings can help construction companies adopt innovative technologies to reduce the number of accidents and injuries on construction sites. The study also identifies areas for future research in this field, highlighting the need for further investigation into the use of these technologies for CSS management.

Compliance with the Use of Personal Protective Equipment (PPE) on Construction Sites in Johor, Malaysia

Article

Jun 2020

The construction environment is one of the most dangerous workplaces where several fatal accidents have been recorded. Personal protective equipment (PPE) has therefore been used as a vital protection in ensuring the health and safety of workers on construction sites. A few findings reveal that PPE is usually not worn by construction workers in Malaysia. The purpose of this research was to investigate the awareness and compliance with PPE usage on construction sites in Johor, Malaysia. A mixed-method research approach was adopted for the study. A total of 95 questionnaires were distributed to workers on nine (9) construction sites in Johor with a response rate of 72%. The data analysis method utilised quantitative analysis using mean values, and qualitative analysis using coding of real-time observations. Both quantitative and qualitative methods reveal head, foot, body, and hand PPE as the most commonly used PPE among construction workers on sites in Johor. The findings also reveal a considerable level of awareness but a low level of compliance with the use of PPE by construction workers in Johor as only three (3) PPE out of eight (8) had a compliance level above the average mean of 2.5. Hence, approaches such as provision of safety training and penalties to workers who fail to comply with PPE usage should be implemented to enhance compliance with the use of PPE on construction sites.

Constructing a Safer Future: Unveiling Health and Safety Challenges in Building Construction Projects in Ethiopia

Article

Full-text available

Mar 2024

The construction industry is one of the most dangerous work sites in the world, having a high rate of accidents, illnesses, and fatalities, construction personnel workplace health and safety are major challenges. The building construction sector in Ethiopia has seen remarkable growth. However, this growth has led to increased health and safety concerns in building construction projects. This study seeks to identify and address health and safety challenges of Ethiopian building construction projects, aiming to enhance the well-being of workers and promote a safer construction environment. This study employed a descriptive study, once reviewing sufficient literature; questionnaires used to collect primary data from the respondent. Based on the findings of the research, the most health and safety challenges are insufficient regulatory enforcement (Mean=4.90), corruption (4.78), inadequate training and awareness (4.24), limited investment in safety equipment (4.11), lack of management commitment (4.09) and resource constraints (3.98). And the research proposed solutions to alleviate the issues these are strengthening regulatory enforcement (Mean=4.37), adopt anti-corruption strategies (4.27), and comprehensive health and safety training programs (4.18) are crucial. These solutions should aim to improve health and safety practices within the building construction projects in the country, contributing to a safer future. In general, the comprehensive insights offer valuable guidance to organizations, policymakers, and regulatory bodies to effectively address health and safety challenges of workers, ultimately fostering a safer construction environment in building construction projects and contributing to sustainable urban growth in Ethiopia.

Constructing a Safer Future: Unveiling Health and Safety Challenges in Building Construction Projects in Ethiopia

Article

Mar 2024

Personal Protective Equipment Detection for Construction Workers: A Novel Dataset and Enhanced YOLOv5 Approach

Article

Full-text available

Jan 2024

Current research on personal protective equipment (PPE) detection has mainly focused on hard hats, overlooking the detection of reflective clothing. Therefore, this study aims to address this research gap comprehensively. We achieve this by creating a novel dataset using semi-automatic labeling techniques and enhancing the YOLOv5 model. The dataset consists of four categories, assessing the presence of hard hats and reflective clothing. Additionally, we introduce an attention mechanism and an improved loss function to tackle challenges related to detecting reflective clothing and overlapping detection frames. Through extensive multi-model comparison experiments, our improved model, AL-YOLOv5, outperforms the baseline model with remarkable advancements of 0.9 AP in the data-limited category and 0.4 mAP overall. Notably, our improved model shows substantial progress in detecting reflective clothing, significantly reducing false detections, and improving overlapping bounding frames. In conclusion, this study contributes to PPE detection accuracy through a novel dataset and an improved model.

Constructing a Safer Future: Unveiling Health and Safety Challenges in Building Construction Projects in Ethiopia

Article

Mar 2024

EJAS Journal of Science and Technology Constructing a Safer Future: Unveiling Health and Safety Challenges in Building Construction Projects in Ethiopia

Article

Jan 2024

Constructing a Safer Future: Unveiling Health and Safety Challenges in Building Construction Projects in Ethiopia

Article

Jan 2024

Lightweight construction safety behavior detection model based on improved YOLOv8

Article

Full-text available

Apr 2025

This paper studies a lightweight construction safety behavior detection model based on improved YOLOv8, aiming to improve the detection accuracy of safety behaviors on construction sites and achieve lightweight models. YOLO (You Only Look Once) is an object detection algorithm that can achieve real-time and efficient object detection by dividing images into grids and predicting the bounding boxes and categories of objects in each grid. Traditional YOLO models often have problems of missed detection and insufficient feature processing when dealing with complex scenes, especially when facing large-scale data sets. In order to solve this problem, this paper improves on the basis of YOLOv8 and uses a lighter Mobilenetv3 as the backbone network to replace the original CSPDARKNET53 to reduce the amount of calculation and improve the processing speed. At the same time, the receptive field is expanded by combining the Receptive Field Block (RFB) module, the ability to capture multi-scale features is enhanced, and the Global Attention Mechanism (GAM)-Attention mechanism is introduced to enhance the recognition ability of local features. Through experimental results, the improved YOLOv8 model performed excellently in detecting five common unsafe behaviors of construction workers, with an mAP of 0.86, a precision of 0.84, a recall rate of 0.87, an F1 value of 0.85, and an IoU of 0.8, which are significantly better than traditional methods. This shows that the model has successfully achieved the goal of lightweight while improving detection accuracy, and has broad application prospects.

An Improved YOLOv5s-Based Algorithm for Unsafe Behavior Detection of Construction Workers in Construction Scenarios

Article

Full-text available

Feb 2025

Currently, the identification of unsafe behaviors among construction workers predominantly relies on manual methods, which are time-consuming, labor intensive, and inefficient. To enhance identification accuracy and ensure real-time performance, this paper proposes an enhanced YOLOv5s framework with three strategic improvements: (1) adoption of the Focal-EIoU loss function to resolve sample imbalance and localization inaccuracies in complex scenarios; (2) integration of the Coordinate Attention (CA) mechanism, which enhances spatial perception through channel-direction feature encoding, outperforming conventional SE blocks in positional sensitivity; and (3) development of a dedicated small-target detection layer to capture critical fine-grained features. Based on the improved model, a method for identifying unsafe behaviors of construction workers is proposed. Validated through a sluice renovation project in Jiangsu Province, the optimized model demonstrates a 3.6% higher recall (reducing missed detections) and a 2.2% mAP improvement over baseline, while maintaining a 42 FPS processing speed. The model effectively identifies unsafe behaviors at water conservancy construction sites, accurately detecting relevant unsafe actions, while meeting real-time performance requirements.

Field Validation of Beacon-Based Indoor Tracking and Localization System for Construction Workers

Article

Sep 2024

Effectiveness of Mobile Applications in Promoting Safety Communication Among Construction Workers

Article

Full-text available

Nov 2024

Construction sites are inherently hazardous environments where effective safety communication is crucial for preventing accidents and injuries. With the widespread adoption of mobile devices, construction companies are exploring mobile applications as a tool to enhance safety communication among workers. This study employed a mixed-methods approach to evaluate the effectiveness of a custom-built mobile safety application at three different construction sites. Site A served as the control group, while sites B and C implemented the app, allowing workers to report hazards, share safety reminders, and receive real-time notifications. Data on reported safety concerns, near-miss incidents, accidents, and worker perceptions were collected over six months. The results demonstrated a significant increase in hazard reporting and awareness of safety issues at sites B and C. Analysis revealed a 28% reduction in accidents at these sites compared to the control site. Surveys and interviews indicated positive perceptions among workers and supervisors, highlighting the app's role in promoting transparency, open communication, and a more collaborative approach to safety management. While acknowledging limitations, such as sample size and potential distractions, the findings suggest mobile applications can be an effective tool for enhancing safety communication and fostering a proactive safety culture in the construction industry. Successful implementation requires addressing challenges like worker distraction, providing adequate training, and encouraging widespread adoption among workers and management.

Assessment of technology adoption in construction safety management: Case study in China

Article

Full-text available

Dec 2024
Sci Progr

The current body of literature indicates that the utilization of emerging digital safety technologies has the potential to enhance the efficacy of safety management in construction significantly. Nevertheless, there has been limited uptake of digital safety technologies within the construction industry in developing countries. This study is focused on evaluating the adoption of digital safety technologies in construction specifically from the perspective of developing countries. Questionnaire data were gathered from project management professionals in China and subjected to analysis using mean ratings, factor analysis, and fuzzy composite evaluation. The developed model for assessing safety technology adoption comprises four primary criterion groups about factors influencing adoption, namely organizational, technological, individual, and external categories. The study results indicate that factors within the technology category exert the most significant influence on the adoption assessment model, followed by those in the external and organizational categories, and finally by individual category factors. These findings contribute to advancing the theory of digital technology adoption in construction safety management research within developing countries. Practitioners can utilize this information to effectively evaluate and compare success rates of safety technology adoption, thereby informing management decisions required for projects. The insights provided by this study will equip practitioners with scientific knowledge for more effective and sustainable program management, ultimately enhancing the success of safety technologies.

Potential of Internet of Things technologies in health and safety management in the Zimbabwean construction industry

Chapter

Full-text available

Dec 2024

YOLOv8s-SNC: An Improved Safety-Helmet-Wearing Detection Algorithm Based on YOLOv8

Article

Full-text available

Dec 2024

The use of safety helmets in industrial settings is crucial for preventing head injuries. However, traditional helmet detection methods often struggle with complex and dynamic environments. To address this challenge, we propose YOLOv8s-SNC, an improved YOLOv8 algorithm for robust helmet detection in industrial scenarios. The proposed method introduces the SPD-Conv module to preserve feature details, the SEResNeXt detection head to enhance feature representation, and the C2f-CA module to improve the model’s ability to capture key information, particularly for small and dense targets. Additionally, a dedicated small object detection layer is integrated to improve detection accuracy for small targets. Experimental results demonstrate the effectiveness of YOLOv8s-SNC. When compared to the original YOLOv8, the enhanced algorithm shows a 2.6% improvement in precision (P), a 7.6% increase in recall (R), a 6.5% enhancement in mAP_0.5, and a 4.1% improvement in mean average precision (mAP). This study contributes a novel solution for industrial safety helmet detection, enhancing worker safety and efficiency.

Digital Twin Enabled Construction Site Monitoring (CSM) Method with Edge-Cloud Collaboration

Conference Paper

Full-text available

Aug 2024

Digital Twin Enabled Collision Early Warning for Smart Site Safety System (SSSS)

Conference Paper

Full-text available

Aug 2024

Real-time Locating System-enabled Digital Twin for Crane Operation Safety Monitoring on Construction Sites

Conference Paper

Full-text available

Aug 2024

Evaluasi penggunaan APD dalam konteks kesehatan dan keselamatan kerja di industri konstruksi

Article

Full-text available

Aug 2024

Permasalahan utama yang diidentifikasi adalah rendahnya kepatuhan penggunaan APD, hanya 54% pekerja yang menggunakan APD secara lengkap, yang disebabkan oleh ketidaknyamanan dan kurangnya pengawasan. Penelitian ini mengkaji penggunaan Alat Pelindung Diri (APD) dalam proyek konstruksi yang sudah dijalankan di Jatinangor, Sumedang, dengan fokus pada tingkat kepatuhan pekerja dan faktor-faktor yang mempengaruhinya. Metode yang digunakan dalam penelitian ini meliputi survei, wawancara, observasi, dan studi kepustakaan. Hasil penelitian menunjukkan bahwa pengawasan yang ketat, edukasi berkelanjutan, penyediaan APD yang ergonomis, dan penerapan insentif serta sanksi tegas dapat meningkatkan kepatuhan penggunaan APD. Kesimpulan dari penelitian ini menekankan pentingnya upaya bersama antara perusahaan dan pengawas untuk memastikan keselamatan kerja melalui penggunaan APD yang konsisten, guna menciptakan lingkungan kerja yang lebih aman dan produktif.

A Novel Three-Stage Collision-Risk Pre-Warning Model for Construction Vehicles and Workers

Article

Full-text available

Jul 2024

Collision accidents involving construction vehicles and workers frequently occur at construction sites. Computer vision (CV) technology presents an efficient solution for collision-risk pre-warning. However, CV-based methods are still relatively rare and need an enhancement of their performance. Therefore, a novel three-stage collision-risk pre-warning model for construction vehicles and workers is proposed in this paper. This model consists of an object-sensing module (OSM), a trajectory prediction module (TPM), and a collision-risk assessment module (CRAM). In the OSM, the YOLOv5 algorithm is applied to identify and locate construction vehicles and workers; meanwhile, the DeepSORT algorithm is applied to the real-time tracking of the construction vehicles and workers. As a result, the historical trajectories of vehicles and workers are sensed. The original coordinates of the data are transformed to common real-world coordinate systems for convenient subsequent data acquisition, comparison, and analysis. Subsequently, the data are provided to a second stage (TPM). In the TPM, the optimized transformer algorithm is used for a real-time trajectory prediction of the construction vehicles and workers. In this paper, we enhance the reliability of the general object detection and trajectory prediction methods in the construction environments. With the assistance afforded by the optimization of the model’s hyperparameters, the prediction horizon is extended, and this gives the workers more time to take preventive measures. Finally, the prediction module indicates the possible trajectories of the vehicles and workers in the future and provides these trajectories to the CRAM. In the CRAM, the worker’s collision-risk level is assessed by a multi-factor-based collision-risk assessment rule, which is innovatively proposed in the present work. The multi-factor-based assessment rule is quantitatively involved in three critical risk factors, i.e., velocity, hazardous zones, and proximity. Experiments are performed within two different construction site scenarios to evaluate the effectiveness of the collision-risk pre-warning model. The research results show that the proposed collision pre-warning model can accurately predict the collision-risk level of workers at construction sites, with good tracking and predicting effect and an efficient collision-risk pre-warning strategy. Compared to the classical models, such as social-GAN and social-LSTM, the transformer-based trajectory prediction model demonstrates a superior accuracy, with an average displacement error of 0.53 m on the construction sites. Additionally, the optimized transformer model is capable of predicting six additional time steps, which equates to approximately 1.8 s. The collision pre-warning model proposed in this paper can help improve the safety of construction vehicles and workers.

Vision-Based Construction Safety Monitoring Utilizing Temporal Analysis to Reduce False Alarms

Article

Full-text available

Jun 2024

Construction safety requires real-time monitoring due to its hazardous nature. Existing vision-based monitoring systems classify each frame to identify safe or unsafe scenes, often triggering false alarms due to object misdetection or false detection, which reduces the overall monitoring system’s performance. To overcome this problem, this research introduces a safety monitoring system that leverages a novel temporal-analysis-based algorithm to reduce false alarms. The proposed system comprises three main modules: object detection, rule compliance, and temporal analysis. The system employs a coordination correlation technique to verify personal protective equipment (PPE), even with partially visible workers, overcoming a common monitoring challenge on job sites. The temporal-analysis module is the key component that evaluates multiple frames within a time window, triggering alarms when the hazard threshold is exceeded, thus reducing false alarms. The experimental results demonstrate 95% accuracy and an F1-score in scene classification, with a notable 2.03% average decrease in false alarms during real-time monitoring across five test videos. This study advances knowledge in safety monitoring by introducing and validating a temporal-analysis-based algorithm. This approach not only improves the reliability of safety-rule-compliance checks but also addresses challenges of misdetection and false alarms, thereby enhancing safety management protocols in hazardous environments.

A State-of-the-art Analysis of Virtual Reality Applications in Construction Health and Safety

Article

Full-text available

Jun 2024

The construction industry contends with high injury rates, emphasizing the need for innovative preventive measures in construction health and safety (CHS). While previous studies have investigated the potential ap plications of virtual reality (VR) in the construction industry for different purposes, an in-depth study on VR in the CHS context is lacking. Hence, this study provides a state-of-the-art analysis of VR applications in CHS, employing a dual scientometric and systematic review approach. A scientometric analysis is conducted to examine annual publication trends, keyword co-occurrences, and science mapping of publication outlets, alongside mapping the contributions of leading countries in this domain. This analysis reveals a marked increase in research interest and identifies central thematic connections within the body of literature. The systematic review assesses VR technologies, including immersive, desktop-based, BIM-based, 3D game-based, and augmented reality, addressing their roles in hazard identification and safety training. The study also underscores challenges like infrastructure, content modeling, and interoperability and proposes directions for future research. Recommendations include probing into VR’s role in cognitive safety risks and the impact of users’ prior safety knowledge on learning outcomes. This study suggests that developing tailored VR experiences for specific user groups could significantly advance safety practices in the construction industry.

Development of an Educational Program for Occupational Health and Safety in Saudi Arabia

Article

Full-text available

May 2024

Creating a comprehensive occupational safety education program is essential to ensuring workers' safety. Occupational health and safety (OHS) education programs are designed to train workers on the proper safety procedures to prevent injuries and illnesses in the workplace. These programs teach workers about the risks associated with different job tasks and guide how to work safely and effectively. Based on expert judgments, this study aims to develop an educational program for OHS in Saudi Arabia. The survey was distributed to the experts in various fields related to OHS and the education sector, and based on their opinions, recommendations were made for establishing OHS education programs in the Universities in Saudi Arabia. The survey findings revealed a strong consensus among experts regarding incorporating OHS education into university curricula. Experts emphasized the need for a balanced approach that combines theoretical knowledge with practical training. They also highlighted the importance of tailoring OHS education programs to the specific needs of different industries and occupations. The results of this study will enable universities to develop effective educational programs for OHS for students and provide graduates with the knowledge and skills necessary to work and promote a culture of safety and prevention of injuries and diseases in the workplace. OHS education is expanding to include sustainability, creating a win-win where safer workplaces also benefit the environment. Educational programs are adapting to teach both OHS and environmental impact, preparing future professionals to champion human well-being.

Artificial intelligence in risk management within the realm of construction projects: A bibliometric analysis and systematic literature review

Article

May 2025

BIM-based automated analysis of dynamic hazards for proactive safety measures during the earthwork construction stage using CCTV data

Article

May 2025
ADV ENG INFORM

GS-LinYOLOv10: A drone-based model for real-time construction site safety monitoring

Article

May 2025

Machine Learning and Internet of Things for Construction Air Pollution Monitoring and Prediction

Chapter

Mar 2025

A three-way decision-based model for occupational risk assessment and classification in the healthcare industry

Article

Apr 2025
APPL SOFT COMPUT

Navigating the shift towards sustainable digital building permits and building logbooks

Article

Mar 2025

The architecture, engineering, construction, and operation sectors face significant sustainability challenges. Environmentally, it contributes significantly to greenhouse gas emissions and resource depletion. Socially, it must address issues such as worker safety and community impact. Economically, the sector struggles to balance cost efficiency with sustainable practices. Digital solutions are expected to support sustainable construction. Digital building permits (DBP) and digital building logbooks (DBL) provide examples of digital solutions that support sustainable construction and building management. DBP and DBL are intertwined to enhance the efficiency and transparency of the construction and building management processes. However, research on how DBP and DBL can address sustainability in practice is limited. To address this research gap, this study uses the UN’s Sustainable Development Goals (SDGs) as an analytical framework for the sustainability of DBP and DBL. The research consisted of four phases. First, an expert group identified and selected the SDGs related to DBP and DBL. Then, the experts identified the relevant targets of the selected SDGs. Subsequently, the expert group specified the DBP and DBL practices that supported the relevant targets. Finally, the expert group organised a workshop with external experts in the study area to verify the practices supporting the SDGs. The study identified DBP and DBL practices contributing to achieving 10 SDGs: 3, 7, 8, 9, 10, 11, 12, 13, 16, and 17. The findings suggest that DBL and DBL practices provide opportunities for environmental, social, and economic sustainability; however, further empirical research is needed. The study concluded that DBP and DBL practices can enhance energy management, reduce carbon emissions, improve resource utilisation, and reduce waste. They also support creating a built environment that is user-friendly and remotely accessible, as well as offering financial benefits and improving efficiency and transparency while minimising errors from human interpretation through automation.

Mp-Mcwt: Message Passing Strategy for Multi-Camera Worker Tracking in Construction

Preprint

Jan 2025

Rebuilding or retrofitting? An assessment of social impacts using Social Life Cycle Assessment

Article

Mar 2025
ENVIRON IMPACT ASSES

In a world that is facing multiple crises, the construction sector occupies a pivotal position due to its significant impacts on the economy, environment, and society. To respond to some of these challenges, the construction industry has implemented sustainability measures, such as decarbonisation and circularity strategies. A noteworthy strategy in this context involves retrofitting a building instead of the traditional approach of demolishing and rebuilding. This alternative not only showcases potential for enhanced energy efficiency and reduced use of raw materials but also presents opportunities for long-term cost savings. Despite evident economic and environmental advantages of these strategies, social impacts have remained relatively unexplored. To address this gap, this study employs, for the first time, a Social Life Cycle Assessment (S-LCA) to systematically evaluate the social implications associated with rebuilding and retrofitting scenarios. It uses a university building in London (United Kingdom) as a case study. Normalisation and uncertainty analysis of the results help interpret the findings. The results of this assessment demonstrate that retrofitting, when compared to rebuilding, provides broad and positive social benefits for various stakeholders involved in the construction process. The findings not only contribute to a more holistic understanding of construction impacts but also offer valuable insights for decision-makers, practitioners, and policymakers, highlighting the importance of considering social dimensions in sustainable construction practices.

Bridging the Competency Gap: Implementing a Hybrid-Collaborative Training Model for CPO Factory Pressure Vessel and Storage Tank Technicians in North Sumatera Province

Conference Paper

Jan 2024

Dynamic hazard analysis on construction sites using knowledge graphs integrated with real-time information

Article

Dec 2024
AUTOMAT CONSTR

Construction, as a significant production activity, is inherently prone to accidents. These accidents often result from a chain of multiple hazards. However, existing methods of hazard analysis are limited to singledimensional network modeling and static analysis, which makes them inadequate for addressing the complexity and variability of construction sites. This paper presents a dynamic construction hazard analysis method that integrates real-time information into knowledge graphs. In this approach, label entities are added to general knowledge graphs, linking hazard entities to their labels. Labels identified through vision-based methods are then incorporated into the graphs, allowing for the effective extraction and updating of subgraphs in response to spatiotemporal changes in the scenario. Additionally, graph analysis metrics have been proposed to evaluate the system from multiple levels. Finally, the method was applied to a bridge foundation construction case, demonstrating its practicality and significance in preventing accidents by enabling dynamic hazard analysis.

Intelligent construction benefits the public: evidence from the opinion analysis on social media

Article

Dec 2024
Eng Construct Architect Manag

Purpose The construction and real estate sectors are vital to national economies, but traditional construction methods often lead to challenges such as safety risks, noise and environmental pollution. While intelligent construction is believed to mitigate these issues, there is a lack of solid empirical evidence on whether it truly benefits the general public. This paper seeks to explore the societal benefits of intelligent construction from the public’s perspective, addressing this research gap. Design/methodology/approach The research adopts a two-step approach. First, topic mining is conducted to identify topics closely related to the public’s daily life, such as environmental impact, construction traffic management and construction technologies. These topics are then analyzed through sentiment analysis using a bidirectional long short-term memory model with attention mechanism to determine whether the public has a favorable view of these aspects of intelligent construction, indirectly demonstrating the benefits to the public. Findings The primary topics identified include “industry development,” “technology enterprise,” “construction equipment,” “intelligent technology,” “environmental protection,” “robots” and “construction traffic management.” Sentiment analysis shows that public sentiment is overwhelmingly positive across all topics and regions, with “environmental protection,” “construction traffic management” and “robots” receiving the most favorable reactions. Originality/value This study provides empirical evidence of the societal benefits of intelligent construction from the public’s viewpoint using social media data. The results highlight the need for continued promotion and adoption of intelligent construction due to its positive impact on society.

Neural topic modeling of machine learning applications in building: Key topics, algorithms, and evolution patterns

Article

Feb 2025
AUTOMAT CONSTR

The application of machine learning (ML) in the building domain has rapidly evolved due to developments in ML algorithms. Abundant studies have reviewed the use of ML algorithms to address building-domain-related challenges, but some research questions remain unclear: (i) what is the landscape of ML application topics in building domain, (ii) what are the preferences among different ML application topics and algorithms, and (iii) how these topics, ML algorithms, and their preferences evolve until forming current landscape. To address these aspects, an ML-based topic modeling (TM) approach was used in this paper to identify all ML application topics, elucidate the horizontal correlation and vertical knowledge hierarchy among the topics to reveal their static correlation and dynamic evolution with ML algorithms. Several findings that answered each research question were drawn, and recommendations that can facilitate balanced and rational ML advancements in the building domain are proposed for future research.

Construction Site Monitoring for Disaster Resilience

Chapter

Dec 2024

Analyzing Occupational Accidents and Exoskeleton Potential in the Construction Industry in Québec, Canada

Article

Dec 2024

Occupational applications: Our analysis of construction firm accident data in Québec, Canada, spanning from January 2019 to June 2022, provides critical insights for ergonomics and human factors practitioners. The predominant accidents involved being struck by objects (31%) and falls (15%), with lacerations and bruises comprising 37% of injuries. Notably, carpenters had the highest accident rate (22%), prompting attention to task-specific safety measures. We also examined musculoskeletal injuries (MSI), finding that bodily reactions (46%) and overexertion (44%) were primary causes. Importantly, we qualitatively explored the potential of exoskeletons as possible proactive safety measures; our results suggested that in 50% of MSI cases, exoskeletons might have helped to mitigate or eliminate risks. These findings underscore the potential for exoskeletons to enhance safety and productivity in the construction industry, offering opportunities for intervention and preventive measures in ergonomics practice. Technical abstract: Background: The construction industry is a hazardous working environment, having a relatively high risk of accidents and injuries compared to other industries. Purpose: We aimed to describe the characteristics of work accidents in a large construction firm in Québec, Canada, using accident data from the Health, Safety, and Environment Department. Methods: Our dataset spanned from January 2019 to June 2022 and included 2065 complete entries for analysis. Accidents were categorized using a standardized classification scheme, augmented with additional accident types to provide more precision. Results: The most common type of accident was being struck by an object, comprising 31% of reported incidents, followed by falls at 15%. Lacerations and bruises were the most prevalent injuries, accounting for 37% of cases, with injuries to the hand (31%) and the head/eye (24%) being the most frequent. Among the trades analyzed, carpenters had the highest accident rate at 22%, closely followed by laborers at 20%. We also focused on occupational musculoskeletal injuries (MSI) in the dataset to qualitatively investigate the viability of exoskeletons as a proactive safety measure. Of the 268 incidents categorized as MSI-related (13% of the 2065 events), bodily reactions (46%) and overexertion (44%) were the leading causes. A detailed qualitative analysis of the event descriptions suggested that if exoskeletons had been made available and used, they could have contributed to reducing or eliminating MSI risk in 50% of cases. Conclusions: These results contribute to enhancing safety and productivity in the construction industry by providing insights into work accidents and task characteristics that can be used to improve exoskeleton design and compatibility with the work to be performed.

Leveraging Semisupervised Learning for Domain Adaptation: Enhancing Safety at Construction Sites through Long-Tailed Object Detection

Article

Full-text available

Nov 2024
J CONSTR ENG M ASCE

Tracking multiple construction workers using pose estimation and feature-assisted re-identification model

Article

Dec 2024
AUTOMAT CONSTR

Peningkatan Kompetensi Keselamatan dan Kesehatan Kerja Tukang Bangunan Kelurahan Sadeng Dengan Pendekatan Partisipatif

Article

Oct 2024

Sadeng Village, Gunungpati, is still facing challenges in implementing Occupational Safety and Health (K3) among builders. Lack of understanding and awareness of the importance of K3 causes risky work practices and has the potential to cause work accidents. This Community Service Program (PPM) aims to increase the competency of builders in implementing K3 principles. The method used is an integrated and participatory approach, including outreach and direct visits to projects and laboratories. The results of activities show increased compliance with K3 principles. It is hoped that this program can be the first step to building a safe and healthy work culture among builders in Sadeng Village, as well as encouraging sustainable implementation of K3 in the future. Keywords: Occupational Safety and Health; K3; Construction Workers; Laboratory; Community Service Abstrak: Kelurahan Sadeng, Gunungpati, masih menghadapi tantangan dalam penerapan Keselamatan dan Kesehatan Kerja (K3) di kalangan tukang bangunan. Kurangnya pemahaman dan kesadaran tentang pentingnya K3 menyebabkan praktik kerja yang berisiko dan berpotensi menimbulkan kecelakaan kerja. Program Pengabdian pada Masyarakat (PPM) ini bertujuan untuk meningkatkan kompetensi tukang bangunan dalam menerapkan prinsip-prinsip K3. Metode yang digunakan adalah pendekatan terpadu dan partisipatif, meliputi sosialisasi dan kunjungan langsung ke proyek dan laboratorium. Hasil kegiatan menunjukkan peningkatan kepatuhan terhadap prinsip-prinsip K3. Diharapkan, program ini dapat menjadi langkah awal untuk membangun budaya kerja yang aman dan sehat di kalangan tukang bangunan di Kelurahan Sadeng, serta mendorong penerapan K3 yang berkelanjutan di masa mendatang.Kata kunci: Keselamatan dan Kesehatan Kerja; K3; Pekerja Konstruksi; Laboratorium; Pengabdian Masyarakat

Priorización de Riesgos en la Gestión de Proyectos de Construcción Incorporando Impactos de Calidad, Seguridad y Sostenibilidad en Perú: Un Enfoque Jerárquico con AHP

Book

Nov 2024

Ante la creciente complejidad y los múltiples desafíos inherentes a los proyectos de construcción, se evidencia la necesidad imperiosa de desarrollar un enfoque más holístico y robusto para la gestión de riesgos. Esta investigación surge con el objetivo de abordar no solo los impactos tradicionales como costo, tiempo, y probabilidad de ocurrencia, sino también de integrar criterios adicionales de calidad, seguridad, y sostenibilidad, factores que han sido históricamente subestimados en los análisis convencionales. En este contexto, la presente investigación se focaliza en la identificación y priorización de los riesgos más críticos asociados a proyectos de construcción, mediante un análisis de expertos de Perú, mediante la aplicación de un proceso analítico jerárquico. El objetivo central es establecer una jerarquización precisa de los riesgos, facilitando la toma de decisiones informada y estratégica.

Scan-vs-BIM Approach for Construction Site Monitoring Optimization

Chapter

Nov 2024

Few-shot object detection and attribute recognition from construction site images for improved field compliance

Article

Nov 2024
AUTOMAT CONSTR

Massive-Scale construction dataset synthesis through Stable Diffusion for Machine learning training

Article

Oct 2024
ADV ENG INFORM

Critical factors affecting workplace well-being at construction sites across countries with different income levels

Article

Aug 2024
Construct Innovat

Purpose This study aims to analyze the critical factors affecting workplace well-being at construction sites across countries with different income levels. Accordingly, this study’s objectives are to identify: critical factors affecting workplace well-being at construction sites in low-, lower-middle-, upper-middle- and high-income countries, overlapping critical factors across countries with different income levels and agreements on the critical factors across countries with different income levels. Design/methodology/approach This study identified 19 factors affecting workplace well-being using a systematic literature review and interviews with construction industry professionals. Subsequently, the factors were inserted into a questionnaire survey and distributed among construction industry professionals across Yemen, Zimbabwe, Malaysia and Saudi Arabia, receiving 110, 169, 335 and 193 responses. The collected data were analyzed using descriptive and inferential statistics, including mean, normalized value, overlap analysis and agreement analysis. Findings This study identified 16 critical factors across all income levels. From those, 3 critical factors overlap across all countries (communication between workers, general safety and health monitoring and timeline of salary payment). Also, 3 critical factors (salary package, working environment and working hours) overlap across low-, low-middle and upper-middle-income countries, and 1 critical factor (project leadership) overlaps across low-middle, upper-middle and high-income countries. The agreements are inclined to be compatible between low- and low-middle-income, and between low- and high-income countries. However, agreements are incompatible across the remaining countries. Practical implications This study can serve as a standard for maintaining satisfactory workplace well-being at construction sites. Originality/value To the best of the authors’ knowledge, this study is the first attempt to analyze factors affecting workplace well-being at construction sites across countries with different income levels.

AI integration in construction safety: Current state, challenges, and future opportunities in text, vision, and audio based applications

Article

Aug 2024
AUTOMAT CONSTR

Multidimensional analysis of supervisors’ safety leadership on safety violations of construction workers: An empirical investigation

Article

Jun 2024
Work

Background: Frontline supervisors have the most frequent interactions with workers on construction projects. Although Supervisors' Safety Leadership (SSL) is commonly practiced, its specific inter-relationship with workers' safety violations remains unclear, especially when it comes to detailed interactions between supervisors and workers, such as supervisors' safety coaching/safety controlling/safety caring against workers' situational/routine safety violations. Objective: This study aims to uncover the intrinsic relationship between SSL and safety violations from the perspective of construction workers with the help of mediating variables at both organizational and individual levels. Methods: A questionnaire survey was conducted to test all hypotheses based on empirical data from 346 construction workers. The path coefficient of the fitted model was then analyzed, including associated mediating effects. Results: Situational safety violations are directly affected only by safety caring (β= -0.161, p < 0.05), while routine safety violations are impacted only by safety coaching (β= -0.159, p < 0.05). SSL can influence different types of safety violations through differing mediators. In particular, safety coaching acts on individuals' routine safety violations mainly through self-efficacy (β= 0.199, p < 0.01; standardized indirect effect = -0.121, 95% CI[-0.226, -0.024]); safety controlling is more oriented to influence individuals' situational safety violations through group safety norm (β= 0.383, p < 0.001; standardized indirect effect = -0.091, 95% CI[-0.177, -0.036]); and safety caring further influences individuals' situational safety violations mainly through safety motivation (β= 0.581, p < 0.001; standardized indirect effect = -0.263, 95% CI[-0.418, -0.146]). Conclusion: The research enhances existing knowledge by clarifying the complex relationships between supervisor behavior and safety outcomes, particularly from the perceptions of construction workers towards supervisors' actions and leadership.

Automated Kinect-based posture evaluation method for work-related musculoskeletal disorders of construction workers

Article

May 2024

Charging Water Load Prediction with a Multilayer Perceptron for an efficient Facility Management and Maintenance of Thermal-Energy Storage Air-Conditioning

Thesis

Full-text available

Apr 2024

Mirza Rayana Sanzana

This research addresses the challenges in Thermal-Energy-Storage-Air-Conditioning (TES-AC) systems by developing a machine learning model for predicting the necessary water volume for chilling. TES-AC technology, utilizing thermal energy storage tanks, offers substantial benefits such as reduced chiller operation, cost savings, and lower carbon emissions. However, determining the optimal chilled water volume poses challenges. The primary objective is to design a machine learning model leveraging Multilayer Perceptron (MLP) for predicting water load, incorporating input variables like weather forecasts, day of the week, and occupancy data. The study validates the impact of weather data on chilled water volume, demonstrating its efficacy in prediction. The MLP-based model is fine-tuned through hyperparameter optimization, achieving a remarkable accuracy of 93.4%. The model provides specific water volume ranges, minimizing errors and aiding facility managers in informed decision-making to minimize disruptions. Technical significance lies in the model's flexibility, allowing retraining for diverse TES-AC plants without requiring specific sensor inputs. This adaptability promotes widespread TES AC adoption, contributing to environmentally friendly practices in building construction. The integration of the model into facility management software enhances predictive capabilities while offering common features, ensuring seamless incorporation into existing systems. The research aligns with Sustainable Development Goals, particularly Goals 11, 12, and 13, emphasizing sustainable cities, responsible consumption, and climate action. By focusing on technical problem-solving, addressing challenges, and emphasizing the social significance through Sustainable Development Goals, this research provides a comprehensive solution to enhance TES-AC efficiency, thereby contributing to greener and more sustainable urban environments.

Automated Hardhat Detection for Construction Safety Applications

Article

Full-text available

Jan 2017

Despite various safety inspections carried out over the years to ensure compliance with regulations and maintain acceptable and safe working conditions, construction is still among the most dangerous industries responsible for a large portion of the total worker fatalities. A construction worker has a chance of 1-in-200 of dying on the job during a 45-year career, mainly due to fires, falls, and being struck or caught-in/between objects. This in part can be attributed to how monitoring the presence and proper use of personal protective equipment (PPE) by safety officers becomes inefficient when surveying large areas and a considerable number of workers. Therefore, this paper takes the initial steps and aims at evaluating existing computer vision techniques, namely object detection methods, in rapidly detecting whether workers are wearing hardhats from images captured on many indoor jobsites. Experiments have been conducted and results highlighted the potential of cascade classifiers, in particular, in accurately and precisely detecting hardhats under various scenarios and for repetitive runs.

Adaptive computer vision-based 2D tracking of workers in complex environments

Article

Full-text available

Jul 2019
AUTOMAT CONSTR

Monitoring of construction workers is important in managing labour productivity. To date, the construction sector relies either on intensive manual observations or intrusive tag-based practices. Visual tracking methods can provide automated and tag-less monitoring. However, no method to date has succeeded in tracking multiple workers, as construction sites are complex environments due to congestion, background clutter and occlusions. In addition, workers have similar appearance and exhibit illumination/scale/posture variations and abrupt changes in movement over the course of their task. To address these shortcomings, we propose a vision-based method that consists of 3 models. Firstly, an adaptive model provides continuous information about the previous position of workers and their appearance features. Secondly, a prediction model is used to calculate the current position of workers, and finally, an appearance model provides accurate localisation. Experimental results show that the proposed method achieves high performance and outperforms the latest relevant state of the art methods.

Enhancing action recognition of construction workers using data-driven scene parsing

Article

Full-text available

Nov 2018

Jun Yang

Vision-based action recognition of construction workers has attracted increasing attention for its diverse applications. Though state-of-the-art performances have been achieved using spatial-temporal features in previous studies, considerable challenges remain in the context of cluttered and dynamic construction sites. Considering that workers actions are closely related to various construction entities, this paper proposes a novel system on enhancing action recognition using semantic information. A data-driven scene parsing method, named label transfer, is adopted to recognize construction entities in the entire scene. A probabilistic model of actions with context is established. Worker actions are first classified using dense trajectories, and then improved by construction object recognition. The experimental results on a comprehensive dataset show that the proposed system outperforms the baseline algorithm by 10.5%. The paper provides a new solution to integrate semantic information globally, other than conventional object detection, which can only depict local context. The proposed system is especially suitable for construction sites, where semantic information is rich from local objects to global surroundings. As compared to other methods using object detection to integrate context information, it is easy to implement, requiring no tedious training or parameter tuning, and is scalable to the number of recognizable objects.

Network Slicing for 5G: Challenges and Opportunities

Article

Full-text available

Aug 2018

Network slicing for 5G is receiving significant attention from the telecommunications industry as a means to provide network as a service (NaaS) for different use cases. Network slicing is a technology which allows network operators to build multiple virtual networks on a shared infrastructure. With network slicing, service providers can deploy their applications and services flexibly and quickly to accommodate specific requirements of diverse services such as augmented reality, online games, e-health and others. As an emerging technology with a number of advantages, network slicing has raised many issues for the industry and academia alike. In this article, we discuss the background and related work in network slicing and propose a framework for 5G network slicing. Finally, we discuss the challenges of network slicing and future research directions. IEEE

Image-and-Skeleton-Based Parameterized Approach to Real-Time Identification of Construction Workers' Unsafe Behaviors

Article

Full-text available

Jun 2018

Workers' unsafe behaviors are one of the main causes for construction accidents. Fully understanding the causes of unsafe behaviors on site will help to prevent them, thus reducing construction accidents. The accurate and timely identification of site workers' unsafe behaviors can aid in the analysis of the causes of unsafe behaviors and prevention of construction accidents. However, the traditional methods (e.g., site observation) of behavior data collection on site is neither efficient nor comprehensive. This paper develops a skeleton-based real-time identification method by combining image-based technologies, construction safety knowledge, and ergonomic theory. The proposed method recognizes unsafe behaviors by simplifying dynamic motions into static postures, which can be described by a few parameters. Three basic modules are involved: an unsafe behavior database, real-time data collection module, and behavior judgement module. A laboratory test demonstrated the feasibility, efficiency, and accuracy of the method. The method has the potential to improve construction safety management by providing comprehensive data for the systematic identification of the causes to workers' unsafe behaviors, such as inappropriate management methods, measures or decisions, personal characteristics, work space and time, as well as warning workers identified as behaving unsafely, if necessary. Thus, this paper contributes to practice and the body of knowledge of construction safety management, as well as research and practice in image-based behavior recognition.

Interaction analysis for vision-based activity identification of earthmoving excavators and dump trucks

Article

Full-text available

Mar 2018
AUTOMAT CONSTR

Activity identification is an essential step to measure and monitor the performance of earthmoving operations. Many vision-based methods that automatically capture and explain activity information from image data have been developed with economic advantages and analysis efficiency. However, the previous methods failed to consider the interactive operations among equipment, and thus limited the applicability to the operation time estimation for productivity analysis. To address the drawback, this research developed a vision-based activity identification framework that incorporates interactive aspects of earthmoving equipment's operation. This framework included four main processes: equipment tracking, action recognition of individual equipment, interaction analysis, and post-processing. The interactions between excavators and dump trucks were examined due to its significant impacts on earthmoving operations. TLD (Tracking-Learning-Detection) was adapted to track the heavy equipment. Spatio-temporal reasoning and image differencing techniques were then implemented to categorize individual actions. Third, interactions were interpreted based on a knowledge-based system that evaluates equipment actions and proximity between operating equipment. Lastly, outliers or noisy results were filtered out considering work continuity. To validate the proposed framework, two experiments were performed: one with the interaction analysis and the other without the analysis. 11,513 image frames from actual earthmoving sites in total were tested. The consequent average precision of activity analysis was enhanced from 75.68% to 91.27% after the interaction analysis was applied. In conclusion, this research contributes to identifying critical elements that explain interactive operations, characterize the vision-based activity identification framework, and improve the applicability of the vision-based method for the automated equipment operations analysis.

Network Slicing for 5G: Challenges and Opportunities

Article

Full-text available

Jan 2017

Network slicing for 5G provides Network-as-a-Service (NaaS) for different use cases, allowing network operators to build multiple virtual networks on a shared infrastructure. With network slicing, service providers can deploy their applications and services flexibly and quickly to accommodate diverse services’ specific requirements. As an emerging technology with a number of advantages, network slicing has raised many issues for the industry and academia alike. Here, the authors discuss this technology’s background and propose a framework. They also discuss remaining challenges and future research directions.

OVIS: ontology video surveillance indexing and retrieval system

Article

Full-text available

Dec 2017

Nowadays, the diversity and large deployment of video recorders result in a large volume of video data, whose effective use requires a video indexing process. However, this process generates a major problem consisting in the semantic gap between the extracted low-level features and the ground truth. The ontology paradigm provides a promising solution to overcome this problem. However, no naming syntax convention has been followed in the concept creation step, which constitutes another problem. In this paper, we have considered these two issues and have developed a full video surveillance ontology following a formal naming syntax convention and semantics that addresses queries of both academic research and industrial applications. In addition, we propose an ontology video surveillance indexing and retrieval system (OVIS) using a set of semantic web rule language (SWRL) rules that bridges the semantic gap problem. Currently, the existing indexing systems are essentially based on low-level features and the ontology paradigm is used only to support this process with representing surveillance domain. In this paper, we developed the OVIS system based on the SWRL rules and the experiments prove that our approach leads to promising results on the top video evaluation benchmarks and also shows new directions for future developments.

Occupational Safety and Health (Osh) Management In Construction Industry: A Review

Article

Full-text available

Aug 2017
INT J OCCUP SAF ERGO

The construction industry plays a significant role in contributing to economy and development globally. During the process of construction, various hazards coupled with the unique nature of the industry contribute to high fatality rates. This review was carried out by referring to previous published studies and related Malaysian legislation documents. Four main elements consisting of human, worksite, management and external elements which cause occupational accidents and illnesses were identified. External and management elements are the underlying causes contributing to occupational safety and health (OSH), while human and worksite elements are more apparent causes of occupational accidents and illnesses. An effective OSH management approach is required to contain all the hazards at construction sites. An approach to OSH management constructed by elements of policy, process, personnel and incentive developed in previous work is explored. Changes on the sub-elements according to previous studies and the related Malaysian legislation are also covered in this review.

Applying Sensor-Based Technology to Improve Construction Safety Management

Article

Full-text available

Aug 2017
SENSORS-BASEL

Construction sites are dynamic and complicated systems. The movement and interaction of people, goods and energy make construction safety management extremely difficult. Due to the ever-increasing amount of information, traditional construction safety management has operated under difficult circumstances. As an effective way to collect, identify and process information, sensor-based technology is deemed to provide new generation of methods for advancing construction safety management. It makes the real-time construction safety management with high efficiency and accuracy a reality and provides a solid foundation for facilitating its modernization, and informatization. Nowadays, various sensor-based technologies have been adopted for construction safety management, including locating sensor-based technology, vision-based sensing and wireless sensor networks. This paper provides a systematic and comprehensive review of previous studies in this field to acknowledge useful findings, identify the research gaps and point out future research directions.

Feasibility of a Drone-Based On-Site Proximity Detection in an Outdoor Construction Site

Conference Paper

Full-text available

Jun 2017

Being struck by equipment, such as a heavy machine and vehicle, is one of the leading causes of occupational fatalities in an outdoor construction project. As demonstrated through previous sensor-based research efforts, the proximity detection of workers and active equipment is useful in identifying potential hazards at construction sites. However, attaching sensors to entities can be burdensome for both contractors and workers. Additionally, the measurable ranges of proximity sensors may not be consistent because of the vulnerability to ambient conditions. Against this backdrop, we propose the use of a drone with computer vision techniques for on-site proximity detection. Specifically, the potential for the visualization of a struck-by hazard on air-view image frames is investigated in this paper. The drone can fly over a construction site so that moving entities are persistently captured with a drone-mounted camera while being tracked by a computer vision-based tracking algorithm (i.e., mean shift embedded particle filter). In addition, single view geometry-based rectification is adopted with the known geometry information of a reference object, to reflect a real scene scale and dynamic distortions derived from the drone’s altitude and location. Based on the rectified proximity within the image plane, a struck-by hazard zone is defined and visualized in the video, which provides intuitive feedback on whether a worker of interest is within the potentially hazardous area or not. To validate the presented method in terms of accuracy and applicability, a site video from a wastewater treatment plant renovation project was collected. The preliminary test results showed that the defined hazard zone on the drone’s image plane can be rectified to reflect reality on the ground plane. The proposed drone-based on-site proximity detection potentially may be used to identify the risk of workers being struck by equipment in an effective way.

Mobile Asset Tracking for Dynamic 3D Crane Workspace Generation in Real Time

Conference Paper

Full-text available

Jun 2017

Automatic Detection of Helmet Uses for Construction Safety

Conference Paper

Full-text available

Oct 2016

Health Profile of Construction Workers in Hong Kong

Article

Full-text available

Dec 2016
Int J Environ Res Publ Health

Construction is a manual, heavy, and complex sector concerning the most fatal accidents and high incidence of occupational illnesses and injuries resulting in days away from work. In Hong Kong, “Pilot Medical Examination Scheme for Construction Workers” was launched in 2014 to detect the health problems of their construction workforce. All registered workers under the Construction Workers Registration Board are eligible to join the scheme. The purpose of this paper is to assess the physical condition, physiological status, and musculoskeletal disorders of 942 construction workers in Hong Kong. This study adopted a two-phase design, which includes a basic medical examination to measure the workers’ physiological parameters, such as blood pressure, resting heart rate, glucose, cholesterol, uric acid, liver function test, and renal function test; as well as a face-to-face interview following the medical examination to collect their demographic information and pain experience. Individual characteristics, including gender, age, obesity, alcohol drinking habit, and sleeping habit influenced the health condition of construction workers. Among the participants, 36.1% and 6.5% of them were overweight and obese, respectively. In addition, 43.0%, 38.4%, 16.2%, and 13.9% of the participants exceeded the thresholds of cholesterol, blood pressure, urea nitrogen, and uric urea, correspondingly. Moreover, 41.0% of the participants suffered musculoskeletal pain, where the most frequent painful parts occur in the lower back, shoulder, knees, leg, and neck. Through these findings, a series of important issues that need to be addressed is pointed out in terms of maintaining the physical well-being and reducing musculoskeletal disorders of construction workers. The finding may have implications for formulating proper intervention strategies for the sustainable development of Hong Kong’s construction industry.

SSD: Single Shot MultiBox Detector

Conference Paper

Full-text available

Oct 2016
Lect Notes Comput Sci

We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For

300 \times 300

input, SSD achieves 74.3 % mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for

512 \times 512

input, SSD achieves 76.9 % mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https:// github. com/ weiliu89/ caffe/ tree/ ssd.

Handbook of OSHA Construction Safety and Health

Book

Mar 2006

Mapping Computer Vision Research in Construction: Developments, Knowledge Gaps and Implications for Research

Article

Jul 2019
AUTOMAT CONSTR

Computer vision is transforming processes associated with the engineering and management of construction projects. It can enable the acquisition, processing, analysis of digital images, and the extraction of high-dimensional data from the real world to produce information to improve managerial decision-making. To acquire an understanding of the developments and applications of computer vision research within the field of construction, we performed a detailed bibliometric and scientometric analysis of the normative literature from 2000 to 2018. We identified the primary areas where computer vision has been applied, including defect inspection, safety monitoring, and performance analysis. By performing a mapping exercise, a detailed analysis of the computer vision literature enables the identification of gaps in knowledge, which provides a platform to support future research in this fertile area for construction.

Visual Relationship Detection with Language prior and Softmax

Conference Paper

Dec 2018

Remote proximity monitoring between mobile construction resources using camera-mounted UAVs

Article

Mar 2019
AUTOMAT CONSTR

3D tracking of multiple onsite workers based on stereo vision

Article

Feb 2019
AUTOMAT CONSTR

Varied sensing technologies have been delved in for positioning workers and equipment in construction sites. Vision-based technology has been received growing attentions by the virtue of its tag-free and inexpensive configuration. One of the core research works in this area was the use of stereo camera system for tracking 3D locations of construction resources. However, the previous work was limited to tracking of a single entity. To overcome the limitation, this paper presents a new framework for tracking multiple workers. The proposed framework supplements the previous work by embedding an additional step, entity matching, which finds corresponding matches of tracked workers across two camera views. Entity matching takes advantage of the epipolar geometry and workers' motion directions for finding correct pairs of a worker's projections on two image planes. This paper also presents an effective approach of camera calibration for positioning entities located a few tens of meters away from the cameras. The proposed framework is evaluated based on completeness, continuity, and localization accuracy of the generated trajectories. The evaluation results have shown its capability of retrieving 96% of actual movements, within localization errors of 0.821 m with 99.7% confidence.

Transfer learning and deep convolutional neural networks for safety guardrail detection in 2D images

Article

Oct 2018
AUTOMAT CONSTR

Safety has been a concern for the construction industry for decades. Unsafe conditions and behaviors are considered as the major causes of construction accidents. The current safety inspection of conditions and behaviors heavily rely on human efforts which are limited onsite. To improve the safety performance of the industry, a more efficient approach to identify the unsafe conditions on site is required to supplement the current manual inspection practice. A promising way to supplement the current manual safety inspection is automated and intelligent monitoring/inspection through information and sensing technologies, including localization techniques, environment monitoring, image processing and etc. To assess the potential benefits of contemporary technologies for onsite safety inspection, the authors focused on real-time guardrail detection, as unprotected edges are the ones cause for workers falling from heights. In this paper, the authors developed a safety guardrail detection model based on convolutional neural network (CNN). An augmented data set is generated with the addition of background image to guardrail 3D models and used as training set. Transfer learning is utilized and the Visual Geometry Group architecture with 16 layers (VGG-16) model is adopted to construct the basic features extraction for the neural network. In the CNN implementation, 4000 augmented images were used to train the proposed model, while another 2000 images collected from real construction jobsites and 2000 images from Google were used to validate the proposed model. The proposed CNN-based guardrail detection model obtained a high accuracy of 96.5%. In addition, this study indicates that the synthetic images generated by augment technology can be used to create a large training dataset, and CNN-based image detection algorithm is a promising approach in construction jobsite safety monitoring.

Towards efficient and objective work sampling: Recognizing workers' activities in site surveillance videos with two-stream convolutional networks

Article

Oct 2018
AUTOMAT CONSTR

Capturing the working states of workers on foot allows managers to precisely quantify and benchmark labor productivity, which in turn enables them to evaluate productivity losses and identify causes. Work sampling is a widely used method for this task, while suffers from low efficiency as only one worker is selected for each observation. Attentional selection asymmetry can also bias its uniform object selection assumption. Existing vision-based methods are primarily oriented towards recognizing single, separated activities involving few workers or equipment. In this paper, we introduce an activity recognition method, which receives surveillance videos as input and produces diverse and continuous activity labels of individual workers in the field of view. Convolutional networks are used to recognize activities, which are encoded in spatial and temporal streams. A new fusion strategy is developed to combine the recognition results of the two streams. The experimental results show that our activity recognition method has achieved an average accuracy of 80.5%, which is comparable with the state-of-the-art of activity recognition in the computer vision community, given the severe camera motion and low resolution of site surveillance videos and the marginal inter-class difference and significant intra-class variation of workers' activities. We also demonstrate that our method can underpin the implementation of efficient and objective work sampling. The training and test datasets of the study are publicly available.

Ergonomic posture recognition using 3D view-invariant features from single ordinary camera

Article

Oct 2018
AUTOMAT CONSTR

Manual construction tasks are physically demanding, requiring prolonged awkward postures that can cause pain and injury. Person posture recognition (PPR) is essential in postural ergonomic hazard assessment. This paper proposed an ergonomic posture recognition method using 3D view-invariant features from a single 2D camera that is non-intrusive and widely installed on construction sites. Based on the detected 2D skeletons, view-invariant relative 3D joint position (R3DJP) and joint angle are extracted as classification features by employing a multi-stage convolutional nerual network (CNN) architecture, so that the learned classifier is not sensitive to camera viewpoints. Three posture classifiers regarding arms, back, and legs are trained, so that they can be simultaneously classified in one video frame. The posture recognition accuracies of three body parts are 98.6%, 99.5%, 99.8%, respectively. For generalization ability, the relevant accuracies are 94.9%, 93.9%, 94.6%, respectively. Both the classification accuracy and generalization ability of the method outperform previous vision-based methods in construction. The proposed method enables reliable and accurate postural ergonomic assessment for improving construction workers' safety and healthy.

An intelligent vision-based approach for helmet identification for work safety

Article

Sep 2018
COMPUT IND

Helmets are essential equipments to protect workers from danger during inspection and operation. Considering that some workers would not always obey the regulation, video surveillance systems covering the whole factory and supervisors are needed to monitor whether workers are wearing helmets or not. However, with a large number of surveillance screens, it is difficult to identify any helmet violation behavior during any time, which can lead to severe accidents. With the rapid development of image recognition technologies, computer vision-based inspections have been one of the most important industrial application areas. In this paper, an intelligent vision-based approach for helmet identification is proposed. This approach focuses on monitoring whether workers are wearing helmets or not, at the same time, identifying the colors of helmets. A color-based hybrid descriptor composed of local binary patterns (LBP), hu moment invariants (HMI) and color histograms (CH) is proposed to extract features of helmets with different colors (red, yellow and blue). Then a hierarchical support vector machine (H-SVM) is constructed to classify all features into four classes (red-helmet, yellow-helmet, blue-helmet and non-helmet). This approach is tested on our data set and the average accuracy of helmet identification is 90.3%.

Framework for Location Data Fusion and Pose Estimation of Excavators Using Stereo Vision

Article

Nov 2018

The application of computer vision (CV) in construction projects has been investigated for many years, resulting in several advanced algorithms and methods. However, there is still a need to advance the current methods for improving the productivity of operations and safety on job sites. The excavator is one of the highly used pieces of equipment on construction sites that needs to be monitored to evaluate both safety and productivity. Knowing the productivity of excavators helps to plan the excavation process more accurately. A long queue of trucks waiting for the excavator(s) means paying more money while the trucks are not being loaded. Moreover, excavators have a higher risk of accidents due to their articulated shape compared to other excavation-related equipment. On the other hand, monitoring an object with four degrees of freedom using sensory data is a very difficult task. Therefore, this research investigates the opportunities to fuse CV-based methods and real-time location systems (RTLSs) and apply stereo vision methods to formulate a comprehensive framework for estimating the three-dimensional (3D) poses of excavators as some of the most widely used equipment on construction sites. Instead of using specialized tools, such as off-the-shelf stereo cameras or markers, this study evaluates the applicability of using the surveillance cameras on construction sites as stereo cameras. Moreover, RTLS data and two or more cameras' data are fused by synchronizing the time and coordinate systems of the cameras and RTLS to investigate the potential of enhancing the accuracy of the pose estimation system and reducing the computational load. Finally, the performance of the proposed framework is evaluated by integrating the results of the excavator parts' detection, the backgrounds' subtraction, and the two-dimensional (2D) skeletons' extraction of the parts from each camera's view.

Convolutional Neural Networks: Vision-based Workforce Activity Assessment in Construction

Article

Nov 2018
AUTOMAT CONSTR

Computer vision approaches have been widely used to automatically recognize the activities of workers from videos. While considerable advancements have been made to capture complementary information from still frames, it remains a challenge to obtain motion between them. As a result, this has hindered the ability to conduct real-time monitoring. Considering this challenge, an improved convolutional neural network (CNN) that integrates Red-Green-Blue (RGB), optical flow, and gray stream CNNs, is proposed to accurately monitor and automatically assess workers’ activities associated with installing reinforcement during construction. A database containing photographs of workers installing reinforcement is created from activities undertaken on several construction projects in Wuhan, China. The database is then used to train and test the developed CNN network. Results demonstrate that the developed method can accurately detect the activities of workers. The developed computer vision-based approach can be used by construction managers as a mechanism to assist them to ensure that projects meet pre-determined deliverables.

A Novel Multichannel Internet of Things Based on Dynamic Spectrum Sharing in 5G Communication

Article

Jun 2018

The shortage of spectrum resources has limited the development of Internet of Things (IoT). Fifth generation (5G) network can flexibly support a variety of devices and services, which makes it possible to combine 5G with IoT. In this paper, a novel multichannel IoT is proposed to dynamically share the spectrum with 5G Communication, where an IoT node including transmitter and receiver is designed to perform 5G communication and IoT communication simultaneously. The subchannel sets allocated for 5G communication and IoT communication are defined by two complementary spectrum marker vectors, respectively. Two independent spectrum sequences are generated by calculating the inner product of spectrum marker vector, pseudo-random (PR) phase and power scaling vector. Two time-domain fundamental modulation waveforms (FMW) generated by the inverse fast fourier transform (IFFT) of the spectrum sequences are used to modulate 5G data and IoT data, respectively. The receiver can detect the data using the same spectrum marker vectors as the transmitter. The BER performances of the system using binary modulation and cyclic code shift keying (CCSK) modulation in the cases of spectrum marker error and multiple access are analyzed, respectively. A subchannel and power optimization unit is formulated as a joint optimization problem, which seeks to maximize the 5G throughput under the constraints of minimal IoT throughput, maximal power and maximal interference. An alternative optimization problem is proposed to maximize the IoT throughput while guaranteeing the minimal 5G throughput. A joint optimization algorithm based on Lagrange dual decomposition is proposed to achieve the optimal solution. Simulation results indicate that the proposed IoT can improve the 5G throughput significantly while the IoT throughput is guaranteed.

Matching Construction Workers across Views for Automated 3D Vision Tracking On-Site

Article

Jul 2018

Computer vision-based tracking methods are used to track construction resources for productivity and safety purposes. This type of tracking requires that targets be accurately matched across multiple camera views to obtain a three-dimensional (3D) trajectory out of two or more two-dimensional (2D) trajectories. This matching is straightforward when it involves easily distinguishable targets in uncluttered scenes. This can be challenging in industrial scenes such as construction sites due to congestion, occlusions, and workers in greatly similar high-visibility apparel. This paper proposes a novel vision-based method that addresses all these issues. It uses as input the output of a 2D vision-based tracking method and searches for potential matches in three sequential steps. It terminates only when a positive match is found. The first step returns the strongest candidate by correlating a segment of workers' past 2D trajectories. The second uses geometric restrictions, whereas the third correlates color intensity values. The proposed method features a promising performance of 97% precision, 98% recall, and 95% accuracy.

Automated Detection of Workers and Heavy Equipment on Construction Sites: A Convolutional Neural Network Approach

Article

May 2018
ADV ENG INFORM

Detecting the presence of workers, plant, equipment, and materials (i.e. objects) on sites to improve safety and productivity has formed an integral part of computer vision- based research in construction. Such research has tended to focus on the use of computer vision and pattern recognition approaches that are overly reliant on the manual extraction of features and small datasets (<10k images/label), which can limit inter and intra-class variability. As a result, this hinders their ability to accurately detect objects on construction sites and generalization to different datasets. To address this limitation, an Improved Faster Regions with Convolutional Neural Network Features (IFaster R-CNN) approach is used to automatically detect the presence of objects in real-time is developed, which comprises: (1) the establishment dataset of workers and heavy equipment to train the CNN; (2) extraction of feature maps from images using deep model; (3) extraction of a region proposal from feature maps; and (4) object recognition. To validate the model’s ability to detect objects in real-time, a specific dataset is established to train the IFaster R-CNN models to detect workers and plant (e.g. excavator). The results reveal that the IFaster R- CNN is able to detect the presence of workers and excavators at a high level of accuracy (91% and 95%). The accuracy of the proposed deep learning method exceeds that of current state-of-the-art descriptor methods in detecting target objects on images.

3D reconstruction of a concrete mixer truck for training object detectors

Article

Apr 2018
AUTOMAT CONSTR

Vision-based monitoring methods have been actively studied in the construction industry because they can be used to automatically generate information related to progress, productivity, and safety. Object detection is essentially used in such monitoring methods to infer jobsite context. However, as many classes of construction entities exist in a job site, large amounts of image data are required to train a detection algorithm to detect each class object in images. Although image data augmentation methods using 3D models were proposed, publicly available 3D models are limited to some construction object classes. Therefore, this study proposes a three-dimensional reconstruction method to generate the image data required for training object detectors. To use the generated synthetic images as training data, a histogram of oriented gradient (HOG) descriptor of a target object is obtained from these images. The descriptor is refined by a support vector machine to increase sensitivity to the target object in test images. The performance of the HOG-based object detector is evaluated using real images from ImageNet. The result shows that the proposed method can generate training data more effectively than existing manual data collection practices.

Falls from Heights: A Computer Vision-based Approach for Safety Harness Detection

Article

Feb 2018
AUTOMAT CONSTR

Falls from heights (FFH) are major contributors of injuries and deaths in construction. Yet, despite workers being made aware of the dangers associated with not wearing a safety harness, many forget or purposefully do not wear them when working at heights. To address this problem, this paper develops an automated computer vision-based method that uses two convolutional neural network (CNN) models to determine if workers are wearing their harness when performing tasks while working at heights. The algorithms developed are: (1) a Faster-R-CNN to detect the presence of a worker; and (2) a deep CNN model to identify the harness. A database of photographs of people working at heights was created from activities undertaken on several construction projects in Wuhan, China. The database was then used to test and train the developed networks. The precision and recall rates for the Faster R-CNN were 99% and 95%, and the CNN models 80% and 98%, respectively. The results demonstrate that the developed method can accurately detect workers not wearing their harness. Thus, the computer vision-based approach developed can be used by construction and safety managers as a mechanism to proactively identify unsafe behavior and therefore take immediate action to mitigate the likelihood of a FFH occurring.

Detecting Construction Equipment Using a Region-Based Fully Convolutional Network and Transfer Learning

Article

Mar 2018

For proper construction site management and plan revisions during construction, it is necessary to understand a construction site's status in real time. Many vision-based construction site-monitoring methods exist, but current technology has not achieved the accuracy required to robustly recognize objects such as construction equipment, workers, and materials in actual jobsite images. To address this issue, this paper proposes a deep convolutional network-based construction object-detection method to accurately recognize construction equipment. A deep convolutional network can achieve high performance in various visual tasks, but is not easy to be applied in the construction industry where there is not enough publicly available data for training. This problem is solved by transfer learning, which trains a model for the construction industry by transferring the knowledge of models trained in other domains with a large amount of training data. To evaluate the proposed method, a benchmark data set is created for five classes: a dump truck, excavator, loader, concrete mixer truck, and road roller. This benchmark data set includes various shapes and poses for each class to evaluate the generalization performance of the proposed construction equipment detection model. Experimental results show that the proposed method performs remarkably well, achieving 96.33% mean average precision. In the future, the proposed model can be used to infer the context of construction operations for producing managerial information such as progress, productivity, and safety.

Visual Translation Embedding Network for Visual Relation Detection

Conference Paper

Jul 2017

A Critical Review of Virtual and Augmented Reality (VR/AR) Application in Construction Safety

Article

Nov 2017
AUTOMAT CONSTR

Construction is a high hazard industry which involves many factors that are potentially dangerous to workers. Safety has always been advocated by many construction companies, and they have been working hard to make sure their employees are protected from fatalities and injuries. With the advent of Virtual and Augmented Reality (VR/AR), there has been a witnessed trend of capitalizing on sophisticated immersive VR/AR applications to create forgiving environments for visualizing complex workplace situations, building up risk-preventive knowledge and undergoing training. To better understand the state-of-the-art of VR/AR applications in construction safety (VR/AR-CS) and from which to uncover the related issues and propose possible improvements, this paper starts with a review and synthesis of research evidence for several VR/AR prototypes, products and the related training and evaluation paradigms. Predicated upon a wide range of well-acknowledged scholarly journals, this paper comes up with a generic taxonomy consisting of VR/AR technology characteristics, application domains, safety scenarios and evaluation methods. According to this taxonomy, a number of technical features and types that could be implemented in the context of construction safety enhancement are derived and further elaborated, while significant application domains and trends regarding the VR/AR-CS research are generalized, i.e., hazards recognition and identification, safety training and education, safety instruction and inspection, and so on. Last but not least, this study sets forth a list of gaps derived from the in-depth review and comes up with the prospective research works. It is envisioned that the outcomes of this paper could assist both researchers and industrial practitioners with appreciating the research and practice frontier of VR/AR-CS and soliciting the latest VR/AR applications.

A Deep Hybrid Learning Model to Detect Unsafe Behavior: Integrating Convolution Neural Networks and Long Short-Term Memory

Article

Feb 2018
AUTOMAT CONSTR

Construction Workers Activity Detection Using BOF

Conference Paper

Mar 2017

K. S. Sowmya

Detecting non-hardhat-use by a deep learning method from far-field surveillance videos

Article

Jan 2018
AUTOMAT CONSTR

Hardhats are an important safety measure used to protect construction workers from accidents. However, accidents caused in ignorance of wearing hardhats still occur. In order to strengthen the supervision of construction workers to avoid accidents, automatic non-hardhat-use (NHU) detection technology can play an important role. Existing automatic methods of detecting hardhat avoidance are commonly limited to the detection of objects in near-field surveillance videos. This paper proposes the use of a high precision, high speed and widely applicable Faster R-CNN method to detect construction workers' NHU. To evaluate the performance of Faster R-CNN, more than 100,000 construction worker image frames were randomly selected from the far-field surveillance videos of 25 different construction sites over a period of more than a year. The research analyzed various visual conditions of the construction sites and classified image frames according to their visual conditions. The image frames were input into Faster R-CNN according to different visual categories. The experimental results demonstrate that the high precision, high recall and fast speed of the method can effectively detect construction workers' NHU in different construction site conditions, and can facilitate improved safety inspection and supervision.

Real-Time Video Analytics: The Killer App for Edge Computing

Article

Jan 2017

Video analytics will drive a wide range of applications with great potential to impact society. A geographically distributed architecture of public clouds and edges that extend down to the cameras is the only feasible approach to meeting the strict real-time requirements of large-scale live video analytics.

Image-based construction hazard avoidance system using augmented reality in wearable device

Article

Aug 2017
AUTOMAT CONSTR

Skeleton estimation of excavator by detecting its parts

Article

Oct 2017
AUTOMAT CONSTR

Knowing the near real-time pose of the construction equipment is an important necessity for improving safety and productivity of construction processes. A safer construction site can be achieved by accurately capturing the movements of the equipment and their parts using motion tracking techniques. Recently, the application of Computer Vision (CV) techniques is growing, especially considering that other techniques have difficulties with the deployment and high cost. From the productivity point of view, knowing the pose of the equipment helps to estimate the time that the operator spent on each state of his/her operations. For this purpose, each part or joint of the equipment must be recognized and tracked. Extracting this information requires either using depth sensors or stereo vision, which includes the image information of two or more Red Green Blue (RGB) cameras that have overlapping views. Since the reading range of the common and affordable depth sensors are limited to few meters, this research chooses to use RGB cameras. The most challenging step for this purpose is to process the data of each camera individually for extracting the 2D skeleton of the equipment. After having the 2D skeleton of the equipment from each camera, the 3D pose can be further estimated using the relative rotation and translation information between the cameras coordinate systems. This paper mainly focuses on determining the 2D skeleton of excavators based on the videos received from the cameras available on the site. The method takes advantage of synthetic images of each excavator's part to train the parts' detectors. After detecting the parts, the backgrounds of the detected parts are subtracted. The remaining pixels from the previous step are processed to estimate the skeleton of each part. The final skeleton of the excavator is derived by connecting the individual skeletons of each part to their adjacent parts considering the kinematic information of the excavator.

Potential of Convolutional Neural Network-Based 2D Human Pose Estimation for On-Site Activity Analysis of Construction Workers

Conference Paper

Jun 2017

Vision-based 2D human pose estimation provides a non-invasive and effort-saving means of extracting human motion data to facilitate an automated activity analysis of construction workers, such as unsafe behavior monitoring, ergonomic analysis, and productivity estimation. However, it continues to suffer from inaccuracies and a lack of robustness, particularly under a dynamic and cluttered environment like a construction site where occlusions are prevalent. To address these issues, the authors apply convolutional neural network (CNN) to human detection and pose estimation on sequential images from site conditions. Using the benchmark training datasets that do not include any images taken from the site, the result of 2D pose estimation in testing data shows that this approach achieves a high level of accuracy and robustness considering the presence of partial occlusion. The potential of this human pose estimation method a under dynamic and cluttered construction environment is demonstrated, and its further applications for a worker activity analysis are discussed.

Adaptive Detector and Tracker on Construction Sites Using Functional Integration and Online Learning

Article

Mar 2017

Tracking construction equipment is a major task when monitoring work in progress and performance on construction sites. Real-time location data of heavy equipment can be used not only to prevent collision accidents but also to predict work types and idle time. Many researchers have investigated the two-dimensional (2D) tracking of construction equipment from images. However, this method still frequently fails to track construction equipment in the long term due to the high interclass/intraclass variations of construction equipment and sites. In order to overcome this problem, this paper adapts and customizes a tracking method composed of two main concepts for (1) functional integration of a detector and a tracker and (2) real-time online learning using an automatically developed training database on site. The functional integration is first used to solve retracking issues and provide information used for database development. On the other hand, the online learning focuses on the use of a detector, which utilizes a site-customized database that is developed and updated automatically in real time. Validation was conducted using video stream data collected from four different construction sites. The average precision, recall rates, and data sampling accuracy were 86.53, 86.21, and 79.35%, respectively. The eigenvalues were also calculated as 0.66 and 0.39. The experiment results show the proposed method is able to consider the diverse characteristics of construction equipment and sites with promising performance. The contribution of this study is to improve performance and applicability of the functional integration and online learning for enhancing site awareness in the construction domain.

Article

Dec 2016
OCCUP ENVIRON MED

Objectives: Examine trends and patterns of work-related musculoskeletal disorders (WMSDs) among construction workers in the USA, with an emphasis on older workers. Methods: WMSDs were identified from the 1992-2014 Survey of Occupational Injuries and Illnesses (SOII), and employment was estimated from the Current Population Survey (CPS). Risk of WMSDs was measured by number of WMSDs per 10 000 full-time equivalent workers and stratified by major demographic and employment subgroups. Time series analysis was performed to examine the trend of WMSDs in construction. Results: The number of WMSDs significantly dropped in the US construction industry, following the overall injury trends. However, the rate of WMSDs in construction remained higher than in all industries combined; the median days away from work increased from 8 days in 1992 to 13 days in 2014, and the proportion of WMSDs for construction workers aged 55 to 64 years almost doubled. By occupation, construction labourers had the largest number of WMSD cases, while helpers, heating and air-conditioning mechanics, cement masons and sheet metal workers had the highest rates of WMSDs. The major cause of WMSDs in construction was overexertion, and back injuries accounted for more than 40% of WMSDs among construction workers. The estimated wage loss for private wage-and-salary construction workers was $46 million in 2014. Conclusions: Construction workers continue to face a higher risk of WMSDs. Ergonomic solutions that reduce overexertion-the primary exposure for WMSDs-should be adopted extensively at construction sites, particularly for workers with a higher risk of WMSDs.

Learning Multi-domain Convolutional Neural Networks for Visual Tracking

Conference Paper

Jun 2016

You Only Look Once: Unified, Real-Time Object Detection

Conference Paper

Jun 2016

Learning to Track at 100 FPS with Deep Regression Networks

Conference Paper

Oct 2016
Lect Notes Comput Sci

Machine learning techniques are often used in computer vision due to their ability to leverage large amounts of training data to improve performance. Unfortunately, most generic object trackers are still trained from scratch online and do not benefit from the large number of videos that are readily available for offline training. We propose a method for offline training of neural networks that can track novel objects at test-time at 100 fps. Our tracker is significantly faster than previous methods that use neural networks for tracking, which are typically very slow to run and not practical for real-time applications. Our tracker uses a simple feed-forward network with no online training required. The tracker learns a generic relationship between object motion and appearance and can be used to track novel objects that do not appear in the training set. We test our network on a standard tracking benchmark to demonstrate our tracker’s state-of-the-art performance. Further, our performance improves as we add more videos to our offline training set. To the best of our knowledge, our tracker (Our tracker is available at http:// davheld. github. io/ GOTURN/ GOTURN. html) is the first neural-network tracker that learns to track generic objects at 100 fps.

Continuous localization of construction workers via integration of detection and tracking

Article

Dec 2016
AUTOMAT CONSTR

Tracking construction workers onsite yields trajectory data that can be very useful in safety and productivity monitoring. Tracking via cameras and computer vision-based methods is one of the tracking options available and deemed suitable for worker tracking in congested outdoor environments. However, this option has yet to yield a practical, robust solution, because of its poor performance in handling occlusions and maintaining trajectories over time. This is due, on one hand, to the tracking methods' degeneracy issues, and the use of detection methods only for initializing trackers. On the other hand, detection methods alone are unable to yield robust trajectories because they treat each frame independently and do not account for temporal effects. Furthermore, no detection method satisfies high performance requirements in both precision and recall. Higher precision entails lower recall. False positives or false negatives limit practical applications. To resolve these issues, this paper presents a hybrid method for locating construction workers that fuses tracking and detection together throughout the tracking process. The detection method is designed to hit higher precision while the tracking method supplements the reduced lower recall. Tracking allows entity matching across frames producing trajectories for all identified workers, while tracking of an occluded worker is automatically terminated and resumed when the occlusion gets cleared. Experimental tests on onsite videos demonstrate that the proposed combination improves overall tracking performance.

Mapping Human Resource Management: Reviewing the field and charting future directions

Article

Sep 2017
Hum Resource Manag Rev

Using recent advances in science mapping, this article systematically reviews the Human Resource Management (HRM) field. We analyze 12,157 HRM research articles published over 23 years to reveal the topic content and intellectual structure of HRM scholarship. A downloadable, searchable HRM topic map is provided (http://bit.ly/HR-Map) that reveals: a) 1702 HRM article topics, b) the number of articles on each topic, c) topic relations, trends, and impact, and d) five major HRM topic clusters. We discuss the overall intellectual structure of HRM scholarship and review the five topic clusters. Next, the topic content of HRM scholarship is compared to that of 6114 articles from the practitioner-oriented outlet HR Magazine. We identify 100 topics emphasized to a much greater degree in the practitioner-oriented literature. Seven key themes for future research that could help align HRM scholarship with the interests of HR practitioners are identified and discussed.

Tracking-based 3D human skeleton extraction from stereo video camera toward an on-site safety and ergonomic analysis

Article

Jul 2016
Construct Innovat

Purpose As a means of data acquisition for the situation awareness, computer vision-based motion capture technologies have increased the potential to observe and assess manual activities for the prevention of accidents and injuries in construction. This study thus aims to present a computationally efficient and robust method of human motion data capture for the on-site motion sensing and analysis. Design/methodology/approach This study investigated a tracking approach to three-dimensional (3D) human skeleton extraction from stereo video streams. Instead of detecting body joints on each image, the proposed method tracks locations of the body joints over all the successive frames by learning from the initialized body posture. The corresponding body joints to the ones tracked are then identified and matched on the image sequences from the other lens and reconstructed in a 3D space through triangulation to build 3D skeleton models. For validation, a lab test is conducted to evaluate the accuracy and working ranges of the proposed method, respectively. Findings Results of the test reveal that the tracking approach produces accurate outcomes at a distance, with nearly real-time computational processing, and can be potentially used for site data collection. Thus, the proposed approach has a potential for various field analyses for construction workers’ safety and ergonomics. Originality/value Recently, motion capture technologies have rapidly been developed and studied in construction. However, existing sensing technologies are not yet readily applicable to construction environments. This study explores two smartphones as stereo cameras as a potentially suitable means of data collection in construction for the less operational constrains (e.g. no on-body sensor required, less sensitivity to sunlight, and flexible ranges of operations).

Vision-Based Excavator Detection and Tracking Using Hybrid Kinematic Shapes and Key Nodes

Article

Jun 2016

Enhancing workplace safety continues to be a major task in the construction industry. Approximately 75% of struck-by fatalities are caused by inappropriate spatial-temporal relationships between construction workers and heavy equipment. Construction safety can be improved if the location and movement of heavy equipment are tracked in real time. However, detecting and tracking heavy equipment with kinematic joints and changing poses, such as excavators, is still a challenge for vision-based sensing methods. This study proposes to detect and track excavators using stereo cameras based on hybrid kinematic shape and key node features. Specifically, templates of excavator components are synthesized for detection following kinematic constraints of each component. Thereafter, a fast directional chamfer matching algorithm is used to detect the excavator components, and the detected components are articulated at the key nodes. Finally, the three-dimensional positions of the key nodes are tracked through triangulation to depict the excavator movements. Results from field experiments demonstrated that concatenating the detected components following a matching order enhances the detection performance. It is also found that the stereo triangulation enables efficient tracking of excavator movements by targeting at the key nodes.

A critical review of vision-based occupational health and safety monitoring of construction site workers

Abstract

No full-text available

Recommended publications

Improving River Health Through Mitigation and Monitoring

Helicopter HUM/FDR: Benefits and developments

Helicopter HUM/FDR: Benefits and Developments