Automated defect detection in clay printing

Additive manufacturing (AM) of eco-friendly materials has the potential to decarbonize the construction industry by enabling the creation of complex structures with minimal waste. Clay has been integrated into AM processes as a building material, giving rise to an emerging research field referred to as “clay printing”. Defects, such as tearing and sagging, are common in clay printing and could affect the structural integrity, load-bearing capacity, and overall durability of the structures. However, limited research on defect detection in clay printing and lack of datasets restrict the development of defect detection models. This paper presents a tool - the automated defect detection (ADD) preprocessor - developed to generate a dataset for defect detection models in clay printing. The tool uses images and videos as input for preprocessing and labeling images required to build the dataset, meeting the requirements of defect detection models based on convolutional neural networks. The ADD preprocessor is implemented and validated as a proof of concept for clay printing processes. The results demonstrate the capability of the ADD preprocessor to successfully build a dataset for the deployment of defect detection models in clay printing.

Subjects

additive manufacturing

Clay printing

convolutional neural networks

dataset

defect detection

DDC Class

690: Building, Construction

Options

Automated defect detection in clay printing