ML/DL Object Classification and Accuracy Assessment
Train and Infer the class of segmented objects using Machine or Deep Learning methods.
This Add-On introduces new modules and detailed guidance for annotating segmented objects into semantic classes, training classification models, and predicting the class labels of newly segmented objects. Both traditional Machine Learning and Deep Learning workflows are supported.
It also includes modules for quantitatively evaluating segmentation and classification performance by comparing predicted results against corresponding ground-truth annotations.
To facilitate adoption, the archive provides an example dataset and a step-by-step tutorial demonstrating the annotation workflow and practical use of the included modules.
Note: an earlier version was shared featuring only ML method, which may conflict with this new one. In such case, open the Xtra Folder and delete the folder Xtra_MLObjectClassification and restart the software.