UNDERSTANDING AND INTEGRATION OF BATCH NORMALIZATION TO U - NET WITH AN APPLICATION TO SCHEMANTIC SEGMENTATION OF AERIAL IMAGES

This repository showcases the integration of batch normalization into a customized U-Net architecture for semantic segmentation of aerial images. By implementing batch normalization, the model demonstrates improved convergence speed and stability during the training process, leading to enhanced performance in semantic segmentation tasks. The architecture can be scaled up to handle larger datasets and is particularly effective with increased training data.

Overview

Semantic segmentation plays a crucial role in tasks such as object detection, scene understanding, and image classification, especially in aerial imagery analysis. This project focuses on leveraging the U-Net architecture, known for its effectiveness in biomedical image segmentation, and enhancing its performance through the incorporation of batch normalization.

Key Features

• Customized U-Net architecture tailored for semantic segmentation of aerial images.
• Integration of batch normalization layers to improve model convergence and stability.
• Scalability to handle larger datasets and adaptability to diverse aerial image segmentation tasks.
• Enhanced performance with increased training data, leading to more accurate semantic segmentation results.