Skip to main content
. 2024 Oct 27;26(11):911. doi: 10.3390/e26110911

Table 1.

Characterization statistics for the three benchmark datasets.

Dataset Details MIRFLICKR-25K NUS-WIDE MS-COCO
Dataset Size 20,015 186,577 122,218
Training Size 10,000 10,500 10,000
Retrieval Size 18,015 184,477 117,218
Query Size 2000 2100 5000
Number of Categories 24 10 80
Dim of Text Features 1386 1000 2026