A Novel Computer Vision Model for Medicinal Plant Identification Using Log-Gabor Filters and Deep Learning Algorithms

Table 2

Deep learning models.

Reference	Algorithm	Dataset	Accuracy (%)
[66]	GoogLeNet + linear SVM		87.34%.
[67]	Convolution neural network		86%
[68]	Five-layered convolutional neural network (CNN)	Flavia leaf dataset	98.22%.
		Swedish leaf dataset
[69]	CNN-LSTM network with 20 layers		95.06%.
[70]	MobileNetV2		98.97
[71]	Dual-path CNN (DP-CNN)		95.67%
[72]	Dual-path CNN model	14 species of Taiwan’s most prevalent trees	77.1%
[73]	AlexNet, GoogLeNet, VGG-19, ResNet50, and MobileNetV2	Leafsnap image dataset	92.3%
[74]	5-Layer CNN architecture	Flavia leaf dataset Swedish leaf dataset	95.5 98.2
[75]	GoogleNet, VGGNet, and AlexNet	LIFECLEF 2015 dataset	80%
[76]	Two AlexNets pretrained models		99.3%
[77]	ResNet152 and Inception-ResNetv2 architectures with LBP	Swedish leaf dataset	99%
[78]	Seven-layer CNN	Flavia dataset	94%
[79]	AlexNet and GoogLeNet	Flavia	94%; 98%; 99%
		Folio
		Swedish leaf dataset
[80]	17-Layer CNN architecture		97.9%
[81]	VGG19 architecture with a logistic regression classifier	Folio	96%; 96%; 99%,
		Flavia
		Swedish leaf datasets
[82]	AousethNet	Mendeley dataset (MD2020	99%