Researchers from Monash University and Charles Darwin University have developed a new AI system that can identify contaminated construction and demolition wood waste with 91% accuracy. Published in Resources, Conservation and Recycling, the study introduces the first real-world image dataset of contaminated wood waste, dubbed “Contaminated-CDWW.” Under the supervision of Associate Professor Mehrdad Arashpour, the team, led by Ph.D. candidate Madini De Alwis and Dr. Milad Bazli, fine-tuned deep learning models—including convolutional neural networks—to detect six common contamination types using ordinary RGB images.
Contaminated wood—often tainted by paint, chemicals, metals, and other residues—typically ends up in landfills because manual sorting is costly and time-consuming. By deploying the AI system via camera-enabled sorting lines, drones, or handheld devices, recyclers and contractors can automate on-site decision-making, reducing landfill dependency and promoting wood reuse. The study reports strong precision and recall metrics across contamination categories, opening doors to scalable, AI-driven recycling solutions and operational efficiency.
Wood waste accounts for a significant portion of global construction waste, and efficient recycling is vital for advancing circular economy goals. Integrating AI into waste management practices promises to recover valuable resources, cut costs, and support greener construction worldwide, fostering sustainability.
