IJSRP, Volume 9, Issue 6, June 2019 Edition [ISSN 2250-3153]
J.N. Mwero, C.K. Lagat
This paper outlines a study on the effectiveness of deep learning techniques at detecting voids given photographs of concrete surfaces. The proposed deep learning model makes use of a convolutional neural network (CNN) and an artificial neural network (ANN). The model was trained on a dataset of 4,032 images of sizes 32x32 and 128x128, and subsequently obtained a training accuracy of 92.08% and a validation accuracy of 89.08%.