In the dynamic realm of agricultural AI, deep learning-based fruit detection has gained prominence, particularly in smart orchards. These techniques, however, heavily depend on large, manually labeled datasets, a process both time-consuming and labor-intensive.