Developing WasteSAM: A novel approach for accurate construction waste image segmentation to facilitate efficient recycling

Waste Management & Research The Journal for a Sustainable Circular Economy, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 18, 2024

The escalating volume of construction activities and resultant waste generation underscores the imperative for developing sophisticated segmentation models to facilitate efficient sorting recycling processes. This study introduces WasteSAM, an enhanced iteration segment anything model (SAM), specifically tailored address intricate complexities inherent in imagery. Drawing upon a comprehensive dataset comprising over 15,000 masks representing five distinct categories materials, WasteSAM exhibits notably superior capabilities. Quantitative analysis demonstrates significant performance improvements, with outperforming original SAM by average 23.9% dice similarity coefficient 30.0% normalized surface distance metrics. integration stereo-image techniques refining training has facilitated more accurately discerning three-dimensional structure thereby augmenting precision classification. Noteworthy is model’s adeptness handling textures patterns across diverse imaging modalities, including varying lighting conditions complex object interactions. While showing promising results, this also highlights need high-quality, datasets that reflect real-world site complexities, rather than merely larger datasets.

Language: Английский

Automatic recognition of construction waste based on unmanned aerial vehicle images and deep learning

Journal of Material Cycles and Waste Management, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 10, 2024

Language: Английский

The development of a waste management and classification system based on deep learning and Internet of Things

Environmental Monitoring and Assessment, Journal Year: 2024, Volume and Issue: 197(1)

Published: Dec. 26, 2024

Language: Английский