About This Project
A deep learning tool for segmenting skin lesions in dermoscopic images, built on a U-Net architecture trained in PyTorch. Users upload an image through an interactive Streamlit app and see the predicted mask alongside Dice and IoU performance metrics.
Key Features
- U-Net segmentation in PyTorch
- Dice and IoU evaluation
- Interactive Streamlit deployment
Technologies
PyTorchComputer VisionMedical AIPython