A Cross-Modal Image and Text Retrieval Method Based on Efficient Feature Extraction and Interactive Learning CAE

Table 2

MAP (R = 50) values of different methods on three datasets.

Datasets	MAP values
	Methods	Image query	Text query	Average
Flickr30K	Reference [15]	0.215	0.237	0.226
	Reference [28]	0.304	0.312	0.328
	Reference [22]	0.281	0.335	0.308
	The proposed method	0.338	0.379	0.359
MSCOCO	Reference [15]	0.198	0.264	0.231
	Reference [28]	0.293	0.319	0.301
	Reference [22]	0.275	0.318	0.297
	The proposed method	0.324	0.343	0.334
Pascal VOC 2007	Reference [15]	0.192	0.198	0.195
	Reference [28]	0.279	0.295	0.262
	Reference [22]	0.251	0.247	0.249
	The proposed method	0.306	0.311	0.309