Ship Segmentation U-Net Model
This repository hosts a pre-trained U-Net model for ship segmentation in satellite images. U-Net is a powerful convolutional neural network architecture designed for image segmentation tasks, making it well-suited for detecting and delineating ship boundaries in aerial imagery.
Model Architecture
The U-Net model architecture consists of two main parts: the encoder (contracting path) and the decoder (expanding path).
Encoder (Contracting Path)
Convolutional Layers: The encoder uses a series of convolutional layers to capture spatial features. Each convolutional block is followed by a ReLU activation function.
Pooling Layers: Max pooling operations are applied to reduce the spatial dimensions progressively, allowing the network to learn hierarchical features.
Decoder (Expanding Path)
Up-Convolutional Layers: The decoder path uses transposed convolutions to upsample the feature maps.
Skip Connections: Feature maps from the encoder are concatenated with the upsampled feature maps from the decoder. These skip connections help preserve spatial information and improve segmentation accuracy.
Final Convolution: A 1x1 convolution is applied at the end to produce the segmentation map, where each pixel is assigned a class label.
