Researchers have developed a new AI system that can automatically identify contaminated construction and demolition wood waste.
The teams from Monash University and Charles Darwin University (CDU) trained and tested deep-learning models that detect contaminated wood using images.
“This new system could be deployed via camera-enabled sorting lines, drones or handheld tools to support on-site decision-making,” explains Madini De Alwis who led the research team and is a PhD candidate at Monash’s Department of Civil and Environmental Engineering.
The team found the tool was able to identify six major types of wood contamination including asbestos, creosote, fungi, metal, mould and mildew, and painted wood.
“We curated the first real-world image dataset of contaminated construction and demolition wood waste,” says De Alwis.
Using technology to reduce waste
Typically, contaminated wood from construction and demolition sites ends up in landfills due to the difficulty of sorting it manually.
Wood waste is reported to be one of the largest components of construction waste globally. In Australia, around 1.306 million tonnes of wood are dumped in landfills annually.
Most of it can be recycled, but contamination from paint, chemicals, metals and other construction residues makes sorting difficult and costly.
Computer vision technology has been explored in general waste streams, but its application to contaminated wood waste has remained limited.
“By fine-tuning state-of-the-art deep learning models, including CNNs and Transformers, we showed that these tools can automatically recognise contamination types in wood using everyday RGB images,” says Dr Milad Bazli from CDU who led the project alongside De Alwis.
De Alwis notes that by enabling automated sorting, the technology could give recyclers and contractors a tool to recover valuable resources and reduce landfill dependency.
Image: Suppplied
Read more: Bioplastics Innovation Hub responds to grave plastic pollution concerns
