This repository consists of a simple convolution neural network that takes in images of surface of buildings and assess if there's structural defects present.
The model is trained using the "Trainsaver.py" file and the "restorer.py" file loads the saved model and an image of a building structure. It then gives a confidence score of defects being present in that image.
This model is built using Tensorflow's low-level matrix operations and gradient optimization methods.