Please select a goal image from training data, for OCC problem. #5
Description
OCITN must select a goal image from images belonging to one class for being an OCC algorithm.
We can apply one of the training data (or its editing/transformations) as a goal image.
However, using outside of one class (ex., Lenna or Jesus) conflicts with learning from one class.
Note that anomaly detection researchers do not have to care about this issue because the goal is to detect anomalies.
This issue is significant for the person who obsesses to learn from one class.