Mangrove Monitoring : Machine Learning

This repo includes all development and tools related to the Machine Learning Development of the Mangrove Monitoring Project

We are working on two related projects: Mangrove Area Estimation and Satellite Super-Resolution.

The Mangrove Area Estimation project involves using machine learning models to identify mangroves from drone imagery. The satellite super-resolution project aims to expand the capabilities of our mangrove identification models by enhancing satellite imagery to allow us to precisely track and identify mangroves anywhere on Earth without needing to deploy drones.

More Info

Current Classification Models:

UNet: ResNet18 Encoder (SSL4EO-12), Mangrove Segmentation Decoder

UNet: DenseNet Encoder (Imagenet), Mangrove Segmentation Decoder

Super-Resolution Models being tested:

Schrödinger Bridge Latent Diffusion

3-layer SRCNN