shulavkarki / LandslideDetection-from-satellite-imagery

Landslide Detection from satellite imagery

Landslide Detection From Satellite Imagery

Using Attention and Autoencoder boosted CNN.
Docs »

View Demo · Report Bug · Request Feature · Send a Pull Request

About The Project

In this project, we carried out 3 experiments for the prediction of landslide in image. Experiments were:

• Pre-trained Model used as fixed extractor and fine tuned with classifier. Different experiment was carried out with different model like ResNet18, Resnet50.
• Autoencoder was used to extract the representaion vector space and trained with out 1D convolution layers. [.....Ongoing]
• Attention was used along with autoencoder and classic CNN. [.......Ongoing]

At last we intend to develop a web application to deploy our trained models.

Built With

• Python
• Pytorch
• Tensorboard

Getting Started

📂 Dataset

Bijie satellite imagery dataset is used for this project.

📤 Application

🚧 Roadmap

See the open issues for a list of proposed features (and known issues).

🔗 Pull Request

This repo is under active development. If you have any improvements / suggestions please file an issue or send in a Pull Request.

📫 Contact

Project Link: Github

👥👥 Team

Akshey Sigdel

Anuskha Gupta

Rabin Pandey

Shulav Karki

🔗 Project Link: Github