1. Introduction

What is the crop-mask project?

The crop-mask project consists of methods and data used for training, evaluating and making regional scale cropland predictions with machine learning models. The project is built using the OpenMapFlow library. The below diagram demonstrates the core utility of crop-mask.

Why is crop-mask needed?

The crop-mask project is a core component of the agricultural monitoring work at NASA Harvest Machine Learning.

How does crop-mask work?

At a high level, crop-mask contributors do at least one of the following:

1. Label satellite time series data (not shown on diagram),
2. Analyze data and run the data pipeline to create machine learning ready datasets,
3. Train machine learning models using new datasets,
4. Create predicted cropland maps using trained models,
5. Analyze and improve predicted cropland maps (not shown in diagram).

Tutorial

OpenMapFlow contains a crop-mask-example mini-project along with a colab notebook tutorial for training a model. The tutorial can be accessed by clicking the above "Open In Colab" button. An accompanying video is available here.