Table 1.

The retrieval accuracy mAP of convolutional neural network as a feature extractor (RL-CNN), bag of visual words (BOVW), and multiple fused global features (MFF) on Malaya–Kew (MK) and University of California Merced (UCM) datasets.

Dataset	Method	mAP (%)
MalayaKew	FE-CNN	88.1%
	BOVW	66.2%
	MFF	52.6%
UCM	FE-CNN	90.5%
	BOVW	86.2%
	MFF	69.8%