A vision-based approach for scaffolding monitoring

Scaffolding represents unavoidable and cost-consuming activities, and the wisely utilization of scaffolding is what the owner seriously concerns in mega construction projects. The efficient and effective monitoring of the scaffolding is important to proper management and safety of construction tasks. This paper illustrates a vision-based approach to promote monitoring of scaffolding during construction. The proposed system utilizes images captured on sites as raw data source and estimates the depth map of the scene. Based on depth differences between scaffolding and backgrounds, scaffolding is recognized and extracted via the segmentation of depth images. By analyzing patterns of scaffolding structures, a model between scaffolding volume and images is established to estimate the productivity of scaffolding tasks. A pilot study is conducted and results show that the proposed approach can efficiently and effectively recognize the scaffolding and it becomes possible to identify its daily progress in order to monitor the productivity.