This project aims to detect cracks in construction elements using thermal images. It leverages deep learning models, specifically the U-Net and SAM models, to identify and segment cracks.
- Crack Detection: Utilizes thermal images to detect and segment cracks in construction materials.
- Model Training: Includes notebooks for training the U-Net and SAM models.
- Evaluation: Provides IoU (Intersection over Union) metrics for model performance.
- Clone the repository:
git clone https://github.com/LeticiaVieirg/crack_thermal_detection.git
- Train models with Train_SAM.ipynb and Train_Unet.ipynb.
- Evaluate models using preprocessed thermal images and calculate IoU.
MIT License