A carregar...
Publicação
Construction site object detection with active transfer learning and weighted adaptive uncertainty-diversity sampling using a small imbalanced dataset
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 9.58 MB | Adobe PDF |
Orientador(es)
Resumo(s)
This paper presents an active learning framework for robust object detection in dynamic construction environments, addressing the challenges of limited labeled data and high annotation costs. The framework integrates YOLOv10 with a weighted, adaptive uncertainty–diversity sampling strategy and employs transfer learning to mitigate cold-start issues and accelerate model convergence. The adaptive fusion mechanism dynamically weights multiple uncertainty measures (classification confidence, class entropy, and bounding box variance) while incorporating sample diversity to prioritize the most informative data. Experiments achieved mAP50 of 0.885 and mAP50–95 of 0.730 by Cycle 8. Rare classes showed notable gains: Circular ducts improved from 0.25 to 0.60 mAP50, and Drywall panels from 0.35 to 0.65. The approach reduced labeling effort by 30–40% compared to random sampling, showing its potential to improve the development of construction element recognition algorithms, including small items that are often not visible, such as safety equipment, building parts, and construction tools.
Descrição
Mannem, K. R., Mengiste, E., Vardhan, D., Soto, B. G. D., & Bação, F. (2026). Construction site object detection with active transfer learning and weighted adaptive uncertainty-diversity sampling using a small imbalanced dataset. Automation In Construction, 183, Article 106819. https://doi.org/10.1016/j.autcon.2026.106819 --- This research benefited from utilizing resources available at the Core Technology Platforms (CTP) of New York University Abu Dhabi (NYUAD). In particular, the algorithms developed in this study used the research computing services at NYUAD's Center for Research Computing and High-Performance Computing (HPC). Different Centers at NYUAD partially supported this research. In particular, the Center for Sand Hazards and Opportunities for Resilience, Energy, and Sustainability (SHORES), funded by Tamkeen under the NYUAD Research Institute Award CG013, the Center for Interacting Urban Networks (CITIES), funded by Tamkeen under the NYUAD Research Institute Award CG001, and the Center for Artificial Intelligence and Robotics (CAIR), funded by Tamkeen under the NYUAD Research Institute Award CG010.This work was also supported by national funds through FCT (Foundation for Science and Technology), under the project UIDB/04152 Information Management Research Centre (MagIC)/NOVA IMS.
Palavras-chave
YOLOv10 Active learning for object detection Uncertainty-diversity sampling Transfer learning NMS-free detection Weighted adaptive uncertainty Small imbalanced dataset Annotation cost reduction Real-time object detection Control and Systems Engineering Civil and Structural Engineering Building and Construction SDG 9 - Industry, Innovation, and Infrastructure SDG 11 - Sustainable Cities and Communities