Expanding continual few-shot learning benchmarks to include recognition of specific instances

Open in new window