State of Art Segmentation Network Architectures

This Repository contains the basic architectural description and implementation of some state of the art Segmentation Networks. Image Segmentation is the process of identifying parts of the image and understanding what object they belong to. It lays the basis for performing object detection and classification.

Table of Contents:

• Datasets
• U-Net Architecture
• Seg-Net Architecture
• Res-Net Architecture
• FCN-8 Architecture

Datasets:

I have used two different datasets to implement these architectures. I have added both the original images and masks with its numpy files.

• Top View Heads Dataset: can be downloaded here
• Lungs Segmentation from chest X-rays Dataset: can be downloaded here

U-Net Architecture:

An illustration of U-NET Architecture

Implementation and Results:

Code: Google Colab Notebook for Lungs Segmentation

Code: Google Colab Notebook for TVH Segmentation

Seg-Net Architecture:

An illustration of Seg-NET Architecture

Implementation and Results:

Code: Google Colab Notebook for TVH Segmentation

Res-Net Architecture:

An illustration of Res-NET Architecture

Implementation and Results:

Code: Google Colab Notebook for TVH Segmentation

FCN-8 Architecture:

An illustration of FCN-8 Architecture

Implementation and Results:

Code: Google Colab Notebook for Lungs Segmentation

Code: Google Colab Notebook for TVH Segmentation