ENVI Machine Learning enables you to quickly perform anomaly detection and supervised and unsupervised classification on a single raster. In addition, you can use Machine Learning to extract data from one or more rasters, generate a training model, and reuse that model to perform anomaly detection or classification on other images. You can also use the ENVI Modeler to create models that will have multiple outputs.

Machine Learning differs from Deep Learning in that it requires less data to train the model since the user performs the manual step of extracting features for it to learn and recognize from the examples. Because of this, ENVI Machine Learning can be useful for classification and anomaly detection for users who do not have a high-performance GPU (a requirement for using ENVI Deep Learning).

For details on the algorithms used in ENVI Machine Learning, see ENVI Machine Learning Algorithms Background.

With ENVI Machine Learning, you can perform classification and anomaly detection on a single raster with the following tools. Click on the links below to learn more about each tool:

• Anomaly Detection

  • Isolation Forest

  • Local Outlier Factor

• Supervised Classification

  • Extra Trees

  • K-Neighbors

  • Linear SVM

  • Naive Bayes

  • Random Forest

  • RBF SVM

• Unsupervised Classification

  • BIRCH

  • Mini Batch K-Means

ENVI Machine Learning also provides these tools, where you can train models on one more rasters, then use the saved models to classify other images using the Machine Learning Classification Tool. Click on the links below to learn more.

• Machine Learning Labeling Tool

• Train Birch Tool

• Train Mini Batch K-Means Tool