Hyuk Soo Cho
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hyuk Soo Cho has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Jongwon Seo
Department of Civil and Environmental Engineering, Hanyang University, Seoul 04763, Republic of Korea
2 shared publications - Abubakar Sharafat
Department of Civil and Environmental Engineering, Hanyang University, Seoul 04763, Republic of Korea
1 shared publication - Kamran Latif
Department of Civil & Environmental Engineering, Hanyang University, Seoul 04763, Republic of Korea
1 shared publication - Muhammad Shoaib Khan
Department of Civil & Environmental Engineering, Hanyang University, Seoul 04763, Korea
1 shared publication
Latest Publications
- Open Access
- 31 July 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (15) , 8505
Multi-Modal Excavator Activity Recognition Using Two-Stream CNN-LSTM with RGB and Point Cloud Inputs
Recently, deep learning algorithms have been increasingly applied in construction for activity recognition, particularly for excavators, to automate processes and enhance safety and productivity through continuous monitoring of earthmoving activities. These deep learning algorithms analyze construction videos to classify excavator activities for earthmoving purposes. However, previous studies have solely focused on single-source external videos, which limits the activity recognition capabilities of the deep learning algorithm. This paper introduces a novel multi-modal deep learning-based methodology for recognizing excavator activities, utilizing multi-stream input data. It processes point clouds and RGB images using the two-stream long short-term memory convolutional neural network (CNN-LSTM) method to extract spatiotemporal features, enabling the recognition of excavator activities. A comprehensive dataset comprising 495,000 video frames of synchronized RGB and point cloud data was collected across multiple construction sites under varying conditions. The dataset encompasses five key excavator activities: Approach, Digging, Dumping, Idle, and Leveling. To assess the effectiveness of the proposed method, the performance of the two-stream CNN-LSTM architecture is compared with that of single-stream CNN-LSTM models on the same RGB and point cloud datasets, separately. The results demonstrate that the proposed multi-stream approach achieved an accuracy of 94.67%, outperforming existing state-of-the-art single-stream models, which achieved 90.67% accuracy for the RGB-based model and 92.00% for the point cloud-based model. These findings underscore the potential of the proposed activity recognition method, making it highly effective for automatic real-time monitoring of excavator activities, thereby laying the groundwork for future integration into digital twin systems for proactive maintenance and intelligent equipment management.
- 31 July 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (15) , 8505
Open AccessMulti-Modal Excavator Activity Recognition Using Two-Stream CNN-LSTM with RGB and Point Cloud Inputs
Recently, deep learning algorithms have been increasingly applied in construction for activity recognition, particularly for excavators, to automate processes and enhance safety and productivity through continuous monitoring of earthmoving activities. These deep learning algorithms analyze construction videos to classify excavator activities for earthmoving purposes. However, previous studies have solely focused on single-source external videos, which limits the activity recognition capabilities of the deep learning algorithm. This paper introduces a novel multi-modal deep learning-based methodology for recognizing excavator activities, utilizing multi-stream input data. It processes point clouds and RGB images using the two-stream long short-term memory convolutional neural network (CNN-LSTM) method to extract spatiotemporal features, enabling the recognition of excavator activities. A comprehensive dataset comprising 495,000 video frames of synchronized RGB and point cloud data was collected across multiple construction sites under varying conditions. The dataset encompasses five key excavator activities: Approach, Digging, Dumping, Idle, and Leveling. To assess the effectiveness of the proposed method, the performance of the two-stream CNN-LSTM architecture is compared with that of single-stream CNN-LSTM models on the same RGB and point cloud datasets, separately. The results demonstrate that the proposed multi-stream approach achieved an accuracy of 94.67%, outperforming existing state-of-the-art single-stream models, which achieved 90.67% accuracy for the RGB-based model and 92.00% for the point cloud-based model. These findings underscore the potential of the proposed activity recognition method, making it highly effective for automatic real-time monitoring of excavator activities, thereby laying the groundwork for future integration into digital twin systems for proactive maintenance and intelligent equipment management.
- Open Access
- 15 July 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (14) , 7899
Voxel-Based Digital Twin Framework for Earthwork Construction
Earthwork construction presents significant challenges due to its unique characteristics, including irregular topography, inhomogeneous geotechnical properties, dynamic operations involving heavy equipment, and continuous terrain updates over time. Existing methods often fail to accurately capture these complexities, support semantic attributes, simulate realistic equipment–environment interactions, and update the model dynamically during construction. Moreover, most current digital solutions lack an integrated framework capable of linking geotechnical semantics with construction progress in a continuously evolving terrain. This study introduces a novel, voxel-based digital twin framework tailored for earthwork construction. Unlike previous studies that relied on surface, mesh, or layer-based representations, our approach leverages semantically enriched voxelization to encode spatial, material, and behavioral attributes at a high resolution. The proposed framework connects the physical and digital representations of the earthwork environment and is structured into five modules. The data acquisition module gathers terrain, geotechnical, design, and construction data. Virtual models are created for the earthwork in as-planned and as-built models. The digital twin core module utilizes voxels to create a realistic earthwork environment that integrates the as-planned and as-built models, facilitating model–equipment interaction and updating models for progress monitoring. The visualization and simulation module enables model–equipment interaction based on evolving as-built conditions. Finally, the monitoring and analysis module provides volumetric progress insights, semantic material information, and excavation tracking. The key innovation of this framework lies in multi-resolution voxel modeling, semantic mapping of geotechnical properties, and supporting dynamic updates during ongoing construction, enabling model–equipment interaction and material-specific construction progress monitoring. The framework is validated through real-world case studies, demonstrating its effectiveness in providing realistic representations, model–equipment interactions, and supporting progress information and operational insights.
- 15 July 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (14) , 7899
Open AccessVoxel-Based Digital Twin Framework for Earthwork Construction
Earthwork construction presents significant challenges due to its unique characteristics, including irregular topography, inhomogeneous geotechnical properties, dynamic operations involving heavy equipment, and continuous terrain updates over time. Existing methods often fail to accurately capture these complexities, support semantic attributes, simulate realistic equipment–environment interactions, and update the model dynamically during construction. Moreover, most current digital solutions lack an integrated framework capable of linking geotechnical semantics with construction progress in a continuously evolving terrain. This study introduces a novel, voxel-based digital twin framework tailored for earthwork construction. Unlike previous studies that relied on surface, mesh, or layer-based representations, our approach leverages semantically enriched voxelization to encode spatial, material, and behavioral attributes at a high resolution. The proposed framework connects the physical and digital representations of the earthwork environment and is structured into five modules. The data acquisition module gathers terrain, geotechnical, design, and construction data. Virtual models are created for the earthwork in as-planned and as-built models. The digital twin core module utilizes voxels to create a realistic earthwork environment that integrates the as-planned and as-built models, facilitating model–equipment interaction and updating models for progress monitoring. The visualization and simulation module enables model–equipment interaction based on evolving as-built conditions. Finally, the monitoring and analysis module provides volumetric progress insights, semantic material information, and excavation tracking. The key innovation of this framework lies in multi-resolution voxel modeling, semantic mapping of geotechnical properties, and supporting dynamic updates during ongoing construction, enabling model–equipment interaction and material-specific construction progress monitoring. The framework is validated through real-world case studies, demonstrating its effectiveness in providing realistic representations, model–equipment interactions, and supporting progress information and operational insights.