Land Segmentation using Satellite Images

Achieved a segmentation accuracy of 85%.

Project Overview

Combined Effcient-Net encoder and CNN to predict forest area using 7000+ TIFF images. Achieved a segmentation accuracy of 85%.

Technologies: Convolutional Neural Networks (CNN), CUDA, Deep Learning, Encoders, Data Science.

Duration: 1 months

Team Size: 6 members

Project Preview

Open Github Repository

Repository Structure

1. Efficient Net.ipynb: Integrates EfficientNet for improved accuracy and performance. Optimized for better performance on larger datasets with fewer parameters.

2. Sequential_CNN_model.ipynb: Implements custom CNN layers for segmentation tasks. Designed for small to mid-sized datasets. Focused on segmenting Earth-related imagery for environmental or geographic applications.

How to Use This Repository

1. Clone the Repository

To get started, clone this repository to your local machine using the following command:

git clone https://github.com/omkoli3114/Earth-Bender.git
cd Earth-Bender

Project Gallery