FusionNetGeoLabel

FusionNetGeoLabel 🌍🛰️

Overview 📚

FusionNetGeoLabel is a cutting-edge deep learning framework tailored for semantic segmentation in remotely sensed imagery. This repository is a culmination of my Ph.D. research, aimed at enhancing the accuracy and efficiency of semantic labeling in satellite and aerial images. 🚀

Author ✍️

Teerapong Panboonyuen
(also known as Kao Panboonyuen)
Ph.D. in Computer Engineering, Chulalongkorn University 🎓

Abstract 📄

Semantic segmentation is a cornerstone in remote sensing, impacting various domains like agriculture 🌾, map updates 🗺️, and navigation 🚗. Despite the prominence of Deep Convolutional Encoder-Decoder (DCED) networks, they often falter in capturing low-level features such as rivers and low vegetation due to architectural constraints and limited domain-specific data.

This dissertation presents an advanced semantic segmentation framework designed for remote sensing, featuring five key innovations:

• Global Convolutional Network (GCN): 🧠 Enhances segmentation accuracy for remote sensing images.
• Channel Attention Mechanism: 🎯 Focuses on the most critical features for better performance.
• Domain-Specific Transfer Learning: 🛠️ Tackles the challenge of limited training data in remote sensing.
• Feature Fusion (FF): 🔄 Effectively integrates low-level features into the model.
• Depthwise Atrous Convolution (DA): 🔍 Refines feature extraction for improved segmentation.

Our experiments on private Landsat-8 datasets and the public “ISPRS Vaihingen” benchmark show that the proposed architecture significantly outperforms baseline models. 📊

Publications & Resources 📚

• Ph.D. Thesis: Semantic Segmentation on Remotely Sensed Images Using Deep Convolutional Encoder-Decoder Neural Network 📜
• Code Repository: GitHub - FusionNetGeoLabel 💻
• Pretrained Models: Download Pretrained Models 📥
• ISPRS Vaihingen Dataset: Download Dataset 🗂️
• ISPRS Vaihingen Leaderboard: Semantic Segmentation Leaderboard 🏆

How to Use 🔧

Training 🏋️

To train the model, follow these steps:

1. Clone the repository:

   git clone https://github.com/kaopanboonyuen/FusionNetGeoLabel.git
   cd FusionNetGeoLabel
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Prepare your dataset and adjust the configuration:
   • Modify settings in config.json as needed.
4. Start training:

   python train.py --config config.json
   

Inference 🔍

To perform inference using a pretrained model:

1. Download the pretrained model:
   • Available here.
2. Run the inference script:

   python inference.py --model path_to_pretrained_model --image path_to_image
   

Citation 📝

If this work contributes to your research, please cite it as follows:

@phdthesis{panboonyuen2019semantic,
  title     = {Semantic segmentation on remotely sensed images using deep convolutional encoder-decoder neural network},
  author    = {Teerapong Panboonyuen},
  year      = {2019},
  school    = {Chulalongkorn University},
  type      = {Ph.D. thesis},
  doi       = {10.58837/CHULA.THE.2019.158},
  address   = {Faculty of Engineering},
  note      = {Doctor of Philosophy}
}

License ⚖️

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

This site is open source. Improve this page.