Few-Shot Detection

Zero-Shot Object Detection

An even more ambitious scenario than few-shot, where the model must detect object classes for which it has received no visual examples, relying solely on semantic descriptions (e.g., attributes or text).

← Tillbaka