Available with Image Analyst license.
The goal of object classification is to determine the class of each feature, such as a building. For example, you can use it to determine if a building is damaged after a natural disaster. Object classification requires the following inputs:
- An input raster that contains the spectral bands
- A feature class that defines the location (for example, an outline or a bounding box) of each feature
You can solve object classification through Convolutional Neural Networks (CNN). There are many CNN-based image classification algorithms. Most algorithms have a backbone that uses CNN architecture, such as Resnet, LeNet-5, AlexNet, or VGG 16, which is then followed by a softmax layer.
Object classification uses the Feature Classifier model type to train a model.