1. Introduction

Remote sensing–based land-cover classification is a cornerstone of modern environmental analysis, agricultural monitoring, and land‑use mapping. In this tutorial, you will work hands‑on with QGIS and two powerful open‑source plugins (AI Segmentation Plugin and the Semi‑Automatic Classification Plugin) to build a complete workflow from raw satellite imagery to an evaluated land‑cover map.

Below you'll find what you'll be able to do after completing this tutorial:

Data Preparation & Environment Setup

  • Create a dedicated QGIS profile for remote sensing workflows.

  • Load and organize reference vector data and backdrop layers.

  • Label and structure ground‑truth points for classification.

Segmentation & Training Data

  • Install and configure the AI Segmentation Plugin.

  • Generate parcel‑level segments and split them into training and test areas.

Image Processing & Classification

  • Download Sentinel‑2 imagery using SCP.

  • Construct a band set and clip imagery to a study area.

  • Import, evaluate, and refine training areas.

  • Set up and run a Random Forest classifier.

  • Preview and produce a full‑scene classification.

Validation & Interpretation

  • Perform an accuracy assessment using test data.

  • Interpret classification outputs and identify potential improvements.