Image classification is a powerful type of image analysis that uses machine learning to identify patterns and differences in land cover in drone, aerial, or satellite imagery. Land cover classification maps can be used to monitor deforestation in vulnerable regions; identify the amount of impervious surfaces on different land parcels for tax and property assessments; create flood maps; aid in watershed monitoring and city planning; and more.
ArcGIS Pro offers a powerful array of tools and options for image classification to help users produce the best results for your specific application.
There are four different classifiers available in ArcGIS: random trees, support vector machine (SVM), ISO cluster, and maximum likelihood. Both supervised and unsupervised classification workflows are supported. In supervised classification, the user identifies classes, then provides training samples of each class for the machine learning algorithm to use when classifying the image. This approach works well when the user has a good understanding of what classes are present in their region of interest or is looking for the presence of specific classes. Unsupervised classification does not require training samples or a given set of classes. Rather, the classifier analyzes the dataset and identifies different statistically significant classes that best fit the dataset. These classes are reported as generic classes and must be assigned to specific categories (such as vegetation or roads) by the user after the classification is complete.
Classification methods also includes pixel-based options, where each individual pixel is assigned a class based solely on the characteristics of that specific pixel, or object-based options, which group neighboring pixels with similar characteristics into segments and classifies the resulting segments rather than individual pixels. There are also tools for assessing accuracy by comparing the classified image to a reference image.
For less experienced users, image classification can be time consuming, complicated, and difficult to produce accurate results. The Image Classification wizard in ArcGIS Pro provides a simplified user experience comprised of best practices to guide beginning (and experienced) users through the classification process.
Explore the following resources to learn more about performing image classification in ArcGIS. (Not sure where to start? Look for the star by Esri's most helpful resources.)
Note:
To use image classification tools or the classification wizard, you'll need ArcGIS Pro with either ArcGIS Image Analyst or ArcGIS Spatial Analyst.ArcGIS help
Review the following links on reference materials for ArcGIS products:
- Get started with image classification with this overview.
- Consult a list of the image segmentation and classification toolset in ArcGIS Pro.
- Learn about the different types of classification methods.
- Learn more about object-oriented image classification.
- Learn more about the Image Classification Wizard in ArcGIS Pro.
ArcGIS blogs, articles, story maps, and technical papers
Review the following supplemental guidance about concepts, software functionality, and workflows:
- Follow along to get hands-on experience with the Image Classification Wizard.*
- Learn how to improve your classification results using spectral profiles.
- Identify impervious surfaces using image classification.
- Read about using spectral unmixing to identify ancient lake deposits.
Videos
Review the following Esri-produced videos that clarify and demonstrate concepts, software functionality, and workflows:
- See how you can identify impervious surfaces with supervised classification tools. (13 mins)
- Watch a technical workshop on image segmentation, classification, and machine learning in ArcGIS Pro. (1 hr)
Training and tutorials
Review the following guided lessons and tutorials based on real-world problems and key ArcGIS skills:
- Walk through exercises to identify and calculate impervious surfaces using image classification. (1.5 hours)*
- Walk through exercises to use land cover classification to measure changes in lakes. (1.25 hours)
- Esri Training's Image Classification Learning Plan includes the following trainings:*
Developer resources
Review the following resources and support for automating and customizing workflows:
- Learn about using the ArcGIS REST API for forest-based classification and regression.
- Learn about using the ArcGIS REST API Raster Analysis service to classify data in your applications.
- Learn about using the ArcGIS API for Python to perform image segmentation and classification.
- Read about how to run a pixel-based classification workflow with the arcgis.learn module.
- Find ArcGIS API for Python code samples and instructions showing how to use supervised classification and deep learning to detect settlements.
Technical support
Review the following troubleshooting resources from Esri's tech support team:
- Learn how to apply a raster function template to symbolize classified data in a mosaic dataset or image service.
Esri Community
Use the online imagery community to connect, collaborate, and share experiences:
- See what the Imagery and Remote Sensing community is saying about image classification.
- What bands can the ArcGIS Pro image classification tools use?
- Learn more about using DSMs or DTMs as additional inputs in your classification.
- Learn how to save training samples in the classification wizard.