Unmanned Aerial Vehicle for Automatic Detection of Concrete Crack using Deep Learning

Cracks in concrete structure can lead to catastrophic damage if not addressed correctly at the right time. Although crack close to ground surfaces are easily addressable, however cracks on the surface of pillars of bridges, high-rise buildings and tall concrete structures are difficult to notice due to their heights and if left unaddressed can be risky and may lead to massive damage. Visual inspection is normally done to identify the cracks but it is time consuming and costly, it is also dangerous for specialists, and improper environments can cause evaluation errors. This paper proposes an efficient method of automatically detecting cracks by using Convolutional Neural Network with the help of an unmanned aerial vehicle. The crack detection model was developed and trained from scratch and achieved a training accuracy of 98.60% and validation accuracy of 96.84% before transferring it to a raspberry pi 4 mounted on an Unmanned Arial Vehicle which was developed and tested with the model.