Bridge Debris Detection Model
Detects debris (debris_model.pt) and water (water_model.pt) accumulation on bridge camera images. Companion model to the bridgemonitoring_tool project.
Model Details
- Architecture: UNet (via segmentation_models_pytorch)
- Input: 256x256 RGB
- Output: binary segmentation mask
- Framework: PyTorch 2.6.0
- Training data: Trained on 160 (160 for the water segmentation model, 160 for the debris segmentation model) manually labeled camera images. Data is under consideration of publication.
Usage
See our bridgemonitoring_tool project.