User-defined data layers can be stacked together. Vision models initially classify subjects in the scene. The validation step becomes a core component to ensure improved detections in future releases once it is grounded by classifications.
The more you personalise your tags, the better the system gets