Federated Learning approach to crop identification from satellite data

Student

Anubha Agrawal

Advisor

Kevin Silverstein

Abstract

Accurate classification of crops is essential for improving agricultural management, supporting sustainable land management, and facing food security challenges. Remote sensing provides a cost-effective approach for agricultural monitoring. The combination of spatial, temporal, and spectral resolution can lead to improvements in the decision-making process. However, data privacy is a growing concern among farmers and many are reluctant to share data or participate in big data communities. How is it possible to allow multiple data owners to collaboratively train and use a shared prediction model while keeping all the local training data private? One answer to this question could be Federated Learning. Federated Learning is a machine learning technique that trains an algorithm across multiple decentralized servers holding local data samples without exchanging them. We aim to answer the following question- Can Federated Learning glean insights from a broader group of clients and combine them to deliver a more effective model of crop classification?

Video

Federated Learning approach to crop identification from satellite data