Transformer-based deep learning model and video dataset for unsafe action identification in construction projects

Article Construction & Building Technology

Ontology-based mapping approach for automatic work packaging in modular construction

Xiao Li et al.

Summary: The study proposes a dynamic ontology-based mapping approach to automatically generate semantic-enriched work packages in modular construction, which significantly improves collaborative management and performance of MC projects.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Review Construction & Building Technology

Natural language processing for smart construction: Current status and future directions

Chengke Wu et al.

Summary: This paper reviews the bottom-level techniques and mainstream applications of NLP in the construction industry over the past two decades, covering core steps and related techniques as well as the application of NLP in information extraction, management, and decision-making. Additionally, the paper identifies four research directions, including improving relation extraction, realizing knowledge base auto-development, integrating multi-modal information, and achieving an accuracy-efficiency trade-off.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Computer Science, Interdisciplinary Applications

3D convolutional neural network-based one-stage model for real-time action detection in video of construction equipment

Seunghoon Jung et al.

Summary: This study proposes a one-stage model called ADVICE based on 3D CNN for real-time action detection in construction equipment videos, utilizing the 3D attention module and feature pyramid network to enhance performance. The evaluation results demonstrate that the proposed method can efficiently detect actions in diverse construction sites, potentially improving safety, productivity, and environmental management in construction projects.

COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING (2022)

添加到收藏夹

Article Construction & Building Technology

Deep learning-based networks for automated recognition and classification of awkward working postures in construction using wearable insole sensor data

Maxwell Fordjour Antwi-Afari et al.

Summary: This study proposes the use of a wearable insole pressure system and recurrent neural network models to automatically recognize and classify different types of awkward working postures in construction. The experimental results show that the GRU model outperforms other deep learning models in terms of accuracy and F1-score. These findings are significant for enhancing the health and safety of construction workers through wearable sensor-based posture recognition and classification.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Construction & Building Technology

Sewer defect detection from 3D point clouds using a transformer-based deep learning model

Yunxiang Zhou et al.

Summary: This research develops a deep learning method called TransPCNet for 3D point cloud defect classification. TransPCNet achieves more accurate and effective results by enhancing feature extraction and learning capability, and introducing a novel loss function to address data imbalance issues.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Automation & Control Systems

Temporal segment graph convolutional networks for skeleton-based action recognition

Chongyang Ding et al.

Summary: This paper proposes a novel temporal segment graph convolutional networks (TS-GCN) for skeleton-based action recognition. The method divides the sequence into several subsequences and applies convolutional networks to capture dynamic information stage by stage, aligning the motion features in temporal domain. Additionally, a graph-adaptive method is introduced to construct individual graphs for each subsequence, increasing the generality of graph construction.

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE (2022)

添加到收藏夹

Article Engineering, Electrical & Electronic

Structured Context Enhancement Network for Mouse Pose Estimation

Feixiang Zhou et al.

Summary: This paper proposes a novel Hourglass network based model for pose estimation of mice, which incorporates effective modules of Structured Context Mixer and Cascaded Multi-level Supervision to enhance the robustness of the network. The proposed approach achieves accurate localization through the use of multi-level prediction information. Experimental results demonstrate its competitive performance against state-of-the-art approaches.

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY (2022)

添加到收藏夹

Article Construction & Building Technology

Assessing exposure to slip, trip, and fall hazards based on abnormal gait patterns predicted from confidence interval estimation

Hoonyong Lee et al.

Summary: This study developed a model to assess workers' exposures to slip, trip, and fall (STF) hazards by predicting abnormal gait patterns. The model was evaluated and validated through a field experiment. It demonstrated that workers' exposures to STF hazards can be continuously and remotely monitored, potentially helping to prevent STFs on construction worksites.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Construction & Building Technology

Automatic concrete crack segmentation model based on transformer

Wenjun Wang et al.

Summary: In this study, a novel SegCrack model for pixel-level crack segmentation using deep learning methods is proposed. The model utilizes a hierarchically structured Transformer encoder to output multiscale features and incorporates a top-down pathway and lateral connections for progressive feature upsampling and fusion. An online hard example mining strategy is also adopted to improve model performance. Experimental results demonstrate SegCrack achieves high precision, recall, F1 score, and mean intersection over union on the test set.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Construction & Building Technology

Real-time mixed reality-based visual warning for construction workforce safety

Shaoze Wu et al.

Summary: This paper integrates Digital Twin, Deep Learning, and Mixed Reality technologies into a real-time visual warning system, which improves workers' safety awareness and risk assessment accuracy, and provides new perspectives for construction safety management.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Review Construction & Building Technology

Deep learning-based data analytics for safety in construction

Jiajing Liu et al.

Summary: This article reviews the application of deep learning in data mining for construction safety management. The review identifies challenges including lack of high-quality database, inadequate deep learning models, and limited application scenarios. The authors suggest future research should focus on the development of dynamic multi-modal knowledge graph and knowledge graph-based decision-making, to promote the practical application of deep learning in construction industry.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Construction & Building Technology

Automated detection of contractual risk clauses from construction specifications using bidirectional encoder representations from transformers (BERT)

Seonghyeon Moon et al.

Summary: This paper presents a clause classification model based on the BERT method for construction specifications, which demonstrates excellent performance in various risk categories. It contributes to improving the review process and risk management in the construction industry.

AUTOMATION IN CONSTRUCTION (2022)

添加到收藏夹

Article Construction & Building Technology

Convolutional neural network modeling strategy for fall-related motion recognition using acceleration features of a scaffolding structure

KangHo Lee et al.

Summary: This study introduces a method based on convolutional neural networks to learn unsafe movement patterns of construction workers using scaffolds' accelerations, aiming to reduce fall-related incidents on construction sites.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Article Construction & Building Technology

Sensor-based computational approach to preventing back injuries in construction workers

Sudip Subedi et al.

Summary: The paper introduces a novel approach to identify the theoretical maximum attainable level of safety, safety frontier, for a given construction task, and demonstrates the process through an actual construction-lab-based case study. By using a depth sensor camera for workers' postural data collection and developing a moment prediction model, the paper aims to provide safety managers with a higher safety benchmark for monitoring construction sites.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Article Construction & Building Technology

Computer vision approaches for detecting missing barricades

Eugene Chian et al.

Summary: The study proposes two computer vision-based detection approaches to automatically detect missing barricades on construction sites, aiming to improve efficiency and reduce labor-intensive tasks. The results show that one of the approaches, MODA, has better performance and several implementation advantages compared to MCA, with an average precision of 57.9% and an average recall of 73.6%. These methods can help site managers take prompt action to mitigate the risks of falls from height accidents.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Review Construction & Building Technology

Computer vision applications in construction: Current state, opportunities & challenges

Suman Paneru et al.

Summary: This article summarizes the applications of computer vision in construction projects, focusing on safety management, progress monitoring, productivity tracking, and quality control, while also discussing the opportunities and challenges that future research needs to address.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Article Construction & Building Technology

Pose guided anchoring for detecting proper use of personal protective equipment

Ruoxin Xiong et al.

Summary: Proper use of personal protective equipment (PPE) is crucial for workplace safety management. The authors introduce a pose-guided anchoring framework for detecting multi-class PPE compliance, which shows high accuracy in detecting hardhat and safety vest. The proposed strategy demonstrates superior detection accuracy and extensibility compared to existing methods.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Article Construction & Building Technology

Monitoring distraction of construction workers caused by noise using a wearable Electroencephalography (EEG) device

Jinjing Ke et al.

Summary: In a construction environment requiring high attention, distraction leads to unsafe behavior and decreased safety performance. This study used EEG to examine the correlation between distraction and brain activity, suggesting that beta and gamma powers in specific brain regions can differentiate between focused and distracted states, providing objective evaluation of sustained attention and attention failures. These indicators can also guide attention training and enhance control over work errors in high-risk workplaces.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Article Construction & Building Technology

ANN-based automated scaffold builder activity recognition through wearable EMG and IMU sensors

Srikanth Sagar Bangaru et al.

Summary: The study proposes an ANN-based automated construction worker activity recognition method that can accurately identify complex construction activities. Through a case study, the proposed methodology shows high accuracy in recognizing scaffold builder activities.

AUTOMATION IN CONSTRUCTION (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Multi-View Mouse Social Behaviour Recognition With Deep Graphic Model

Zheheng Jiang et al.

Summary: This study introduces a novel multi-view model that effectively captures social behaviors of mice from different perspectives, improves feature learning performance, reduces computational costs, and addresses the issue of imbalanced data.

IEEE TRANSACTIONS ON IMAGE PROCESSING (2021)

添加到收藏夹

Article Computer Science, Interdisciplinary Applications