This project aims to detect cracks in concrete surfaces in bridge engineering using a deep learning approach. The solution employs a transfer learning based approach using the VGG16 model as a base, and utilizes an open source dataset. The final layers of the VGG16 model were replaced with a dense layer and an output layer to adapt it to the problem at hand. The model's performance was evaluated using precision, recall, and F1 score as the judging metrics.
The dataset used in this project can be found at the following link: https://cutt.ly/PS_1_dataset
TensorFlow Keras Numpy Scikit-learn
Testing Accuracy is: 0.975
Precision: [1. 0.95238095]
Recall: [0.95 1. ]
F1: [0.97435897 0.97560976]
Confusion matrix:
[[ 95 5]
[ 0 100]]