The classification of construction waste material using a deep convolutional neural network

The management of Construction and Demolition Waste (C&DW) is complex and adds significantly to the overall life cycle cost of projects. On site waste sorting using technologies that automatically identify different materials has the potential to assist in classifying C&DW and reduce costs. The aim of this research was to design and describe a deep convolutional neural network (CNN) to identify 7 typical C&DW classifications (both single and mixed disposal) using digital images of waste deposited in a construction site bin (artefact). This approach emulated authentic construction site scenarios where on-site sorting is difficult. A novel design science methodology was used. The experiments delivered 94% accuracy, classifying both single and mixed C&DW. This accuracy is important on projects where on-site sorting is attempted, as in practice bin contamination escalates project costs and reduces C&DW diversion from landfill. To illustrate potential of the research the innovative artefact is incorporated within a hypothetical case study describing its use in a circular C&DW business model.

Automated summary

The management of Construction and Demolition Waste (C&DW) is complicated and costly. The research developed a deep convolutional neural network to accurately classify C&DW in construction sites, achieving 94% accuracy which is crucial for cost reduction and waste diversion from landfill.