Crack defection detector

The project is an AI model to detect the tiles’ cracks by analyzing the images based on CNN and VGG19. Using a 40000-image, the data set was divided into three sections: 60% for training, 20% for validation, and 20% for test. The images were selected randomly for the test to ensure the model’s accuracy. The model could correctly inform the user about the cracked state of the image as an alert message. The results showed that this solution had great success because the accuracy, precision, and recall reached 96.8%, 98.24%, and 94.98%, respectively, meeting the chosen design requirements achieving the project’s target.

Category: ENGINEERING Country: EGYPT Year: 2021

 


Sarah Mohamed


Sohaila Mohamed