Generalized Vision-Based Detection, Identification and Pose Estimation of Lamps for BIM Integration

Area of Science:

• Computer Vision
• Building Information Modelling (BIM)
• Sensor Technology

Background:

• Accurate building asset management is crucial for operational efficiency.
• Existing methods for lamp detection and localization may lack accuracy or cost-effectiveness.
• Integrating real-time sensor data with BIM requires robust detection algorithms.

Purpose of the Study:

• To develop and validate a computer vision approach for automatic lamp detection, identification, and pose estimation.
• To enhance the integration of sensor-derived lamp data into Building Information Modelling (BIM).
• To create a cost-effective and computationally efficient method for real-time building asset monitoring.

Main Methods:

• Utilizing computer vision algorithms for image and location data processing from low-cost sensors.
• Generalizing lamp detection to various light surface types (polygonal, circular).
• Refining the integration of detected lamp data into Building Information Modelling (BIM) workflows.

Main Results:

• Achieved reliable and accurate real-time detection and pose estimation of lamps.
• Demonstrated successful integration of lamp data into BIM through a case study.
• Validated the method's effectiveness using low computational resources.

Conclusions:

• The developed computer vision methodology provides an accurate, cost-effective, and advanced solution for building lamp management.
• The generalized detection and refined BIM integration enhance the system's applicability.
• The approach is suitable for real-time monitoring and efficient building information management.