Skip to content

TadpoleL/tensorflow-based-semantic-segmentation-models-for-RS

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 

Repository files navigation

TensorFlow-based Semantic Segmentation Models for Remote Sensing

This repository contains various semantic segmentation models implemented using TensorFlow 2.14. The models included are:

  • DeepLab
  • HRNet
  • LSTM
  • MACUNet
  • ResUNet
  • SegNet
  • UNet
  • UNet+++
  • U-HRNet

Features

  • Multispectral Image Support: Both training and prediction scripts support multispectral image data.
  • State-of-the-art Models: Includes popular and state-of-the-art semantic segmentation models tailored for remote sensing applications.

Getting Started

Prerequisites

  • TensorFlow 2.14
  • Python 3.9
  • Additional dependencies listed in requirements.txt

Installation

  1. Clone the repository:

    git clone https://github.com/TadpoleL/tensorflow-based-semantic-segmentation-models-for-RS.git
    cd tensorflow-based-semantic-segmentation-models-for-RS
  2. Install required packages:

    pip install -r requirements.txt

Training

To train a model, run the training script with the appropriate configuration.

Prediction

To perform prediction using a trained model, use the prediction script.

Models Description

  • DeepLab: A deep learning model for semantic image segmentation, offering several variants such as DeepLabV3 and DeepLabV3+.
  • HRNet: High-Resolution Network, which maintains high-resolution representations through the whole process.
  • LSTM: Long Short-Term Memory networks, adapted for segmentation tasks.
  • MACUNet: A variation of UNet with additional attention mechanisms.
  • ResUNet: UNet with residual connections.
  • SegNet: A deep convolutional encoder-decoder architecture for image segmentation.
  • UNet: A convolutional network architecture for fast and precise segmentation of images.
  • UNet+++: An advanced version of UNet with nested and dense skip connections.
  • U-HRNet: Combines the strengths of UNet and HRNet for improved segmentation performance.

Contributing

Contributions are welcome! Please submit pull requests or open issues to contribute to this project.

License

This project is licensed under the MIT License - see the LICENSE file for details.

About

Some common tensorflow2.x based semantic segmentation models for remote sensing that can be applied to multi-band imagery

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages