This repository contains the code and documentation for a deep learning model developed for crack detection. The model achieves excellent performance in crack detection, as indicated by a recall of 1.00, precision of 0.98, and an F-score of 0.99 on the test set.
Ensured security and authentication measures by implementing username and password-based authentication for the dashboard.
Created a web-based dashboard using Gradio to allow users to interact with the model by uploading and predicting on new images
The crack detection model was trained on the PS_1_dataset. This dataset contains a collection of surface crack images used for training, validation, and testing.
- Developed a deep learning model using TensorFlow and Keras.
- Leveraged transfer learning with the InceptionV3 pre-trained model for feature extraction.
- Applied data augmentation techniques to increase model robustness.
- Implemented a web-based dashboard using Gradio for user interaction.
- Ensured security and authentication measures for the dashboard.
- Collaborated with a team to deploy the model on a cloud server.
- Run
problem-1(Surface-Crack).ipynbto train the crack detection model evaluate the model's performance on the test set. - Launch the web-based dashboard:
problem-1 with dashboard(Surface-Crack).ipynb
The model achieved a recall of 1.00, precision of 0.98, and an F-score of 0.99 on the test set, indicating its high robustness and accuracy in crack detection.
For more details about the project, refer to the Final_Crack_Detection_Model.pdf file. It provides a comprehensive overview of the model architecture, training process, and evaluation results.