Monitoring the Progress of Concrete Pouring Activities with Knowledge Graph (KG)-Enhanced Computer

BIM4SmartRegions – Digital Twins for Buildings and Cities presented by LOC and Runder Tisch GIS e.V.

Accurate progress tracking is crucial to minimize delays and waste in concrete pouring. Traditional Computer Vision (CV) struggles to integrate contextual data, limiting accuracy. This research presents a Knowledge Graph (KG)-enhanced CV method. Combining object detection with BIM data, it creates a KG representing spatial-temporal relationships between components and resources (e.g., mixer, bucket, hose, workers). Rule-based logic and Graph Neural Networks (GNN) analyze frames to identify pouring states and cycles. Experiments on ten samples from two sites show 80.3% accuracy (rule-based) and 89.2% (GNN). This KG-enhanced CV approach improves generalizability, offering valuable insights for construction performance, waste reduction, and sustainability.

Click “Unblock content”  to watch the video