Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement
Dataset is an essential factor influencing the accuracy of computer vision (CV) tasks in construction. Although image synthesis methods can automatically generate substantial annotated construction data compared to manual annotation, existing challenges limited the CV task accuracy, such as geometri...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Developments in the Built Environment |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2666165925000109 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832087441383620608 |
|---|---|
| author | Yujie Lu Bo Liu Wei Wei Bo Xiao Zhangding Liu Wensheng Li |
| author_facet | Yujie Lu Bo Liu Wei Wei Bo Xiao Zhangding Liu Wensheng Li |
| author_sort | Yujie Lu |
| collection | DOAJ |
| description | Dataset is an essential factor influencing the accuracy of computer vision (CV) tasks in construction. Although image synthesis methods can automatically generate substantial annotated construction data compared to manual annotation, existing challenges limited the CV task accuracy, such as geometric inconsistency. To efficiently generate high-quality data, a synthesis method of construction data was proposed utilizing Unreal Engine (UE) and PlaceNet. First, the inpainting algorithm was applied to generate pure backgrounds, followed by multi-angle foreground capture within the UE. Then, the Swin Transformer and improved loss functions were integrated into PlaceNet to enhance the feature extraction of construction backgrounds, facilitating object placement accuracy. The generated synthetic dataset achieved a high average accuracy (mAP = 85.2%) in object detection tasks, 2.1% higher than the real dataset. This study offers theoretical and practical insights for synthetic dataset generation in construction, providing a future perspective to enhance CV task performance utilizing image synthesis. |
| format | Article |
| id | doaj-art-eebe5bc23f9340a88142c2d5d646b355 |
| institution | Kabale University |
| issn | 2666-1659 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Developments in the Built Environment |
| spelling | doaj-art-eebe5bc23f9340a88142c2d5d646b3552025-02-06T05:12:48ZengElsevierDevelopments in the Built Environment2666-16592025-03-0121100610Generating synthetic images for construction machinery data augmentation utilizing context-aware object placementYujie Lu0Bo Liu1Wei Wei2Bo Xiao3Zhangding Liu4Wensheng Li5College of Civil Engineering, Tongji University, Shanghai, 200092, China; Key Laboratory of Performance Evolution and Control for Engineering Structures of Ministry of Education, Tongji University, Shanghai, 200092 , ChinaCollege of Civil Engineering, Tongji University, Shanghai, 200092, ChinaCollege of Civil Engineering, Tongji University, Shanghai, 200092, China; Corresponding author.Department of Civil, Environmental, and Geospatial Engineering, Michigan Technological University, Houghton, 49931, Michigan, USASchool of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA, 30332, USAPeking University HSBC Business School, Peking University, 518055, ChinaDataset is an essential factor influencing the accuracy of computer vision (CV) tasks in construction. Although image synthesis methods can automatically generate substantial annotated construction data compared to manual annotation, existing challenges limited the CV task accuracy, such as geometric inconsistency. To efficiently generate high-quality data, a synthesis method of construction data was proposed utilizing Unreal Engine (UE) and PlaceNet. First, the inpainting algorithm was applied to generate pure backgrounds, followed by multi-angle foreground capture within the UE. Then, the Swin Transformer and improved loss functions were integrated into PlaceNet to enhance the feature extraction of construction backgrounds, facilitating object placement accuracy. The generated synthetic dataset achieved a high average accuracy (mAP = 85.2%) in object detection tasks, 2.1% higher than the real dataset. This study offers theoretical and practical insights for synthetic dataset generation in construction, providing a future perspective to enhance CV task performance utilizing image synthesis.http://www.sciencedirect.com/science/article/pii/S2666165925000109Computer visionConstruction sitesObject placementSynthetic dataset generationUnreal Engine (UE) |
| spellingShingle | Yujie Lu Bo Liu Wei Wei Bo Xiao Zhangding Liu Wensheng Li Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement Developments in the Built Environment Computer vision Construction sites Object placement Synthetic dataset generation Unreal Engine (UE) |
| title | Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement |
| title_full | Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement |
| title_fullStr | Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement |
| title_full_unstemmed | Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement |
| title_short | Generating synthetic images for construction machinery data augmentation utilizing context-aware object placement |
| title_sort | generating synthetic images for construction machinery data augmentation utilizing context aware object placement |
| topic | Computer vision Construction sites Object placement Synthetic dataset generation Unreal Engine (UE) |
| url | http://www.sciencedirect.com/science/article/pii/S2666165925000109 |
| work_keys_str_mv | AT yujielu generatingsyntheticimagesforconstructionmachinerydataaugmentationutilizingcontextawareobjectplacement AT boliu generatingsyntheticimagesforconstructionmachinerydataaugmentationutilizingcontextawareobjectplacement AT weiwei generatingsyntheticimagesforconstructionmachinerydataaugmentationutilizingcontextawareobjectplacement AT boxiao generatingsyntheticimagesforconstructionmachinerydataaugmentationutilizingcontextawareobjectplacement AT zhangdingliu generatingsyntheticimagesforconstructionmachinerydataaugmentationutilizingcontextawareobjectplacement AT wenshengli generatingsyntheticimagesforconstructionmachinerydataaugmentationutilizingcontextawareobjectplacement |