Image Segmentation using CNN (U-Net)

Implemented semantic segmentation on:

• DRIVE dataset (medical retinal vessel segmentation, 20 images)
• Oxford Pets dataset (pet foreground segmentation, 7,390 images)

Built an encoder–decoder U-Net architecture with skip connections.

Trained using Dice + BCE loss and evaluated segmentation quality using predicted masks.

Tools used: Python, TensorFlow, Keras, OpenCV, Matplotlib.