. 2021 May 28;7:e572. doi: 10.7717/peerj-cs.572

Table 2. Plant recognition studies based on deep learning.

Researchers	Feature extraction methods	Classification methods	Datasets	Accuracy score (%)
Beikmohammadi & Faez (2018)	MobileNet architecture	Logistic regression classifier	Flavia	99.60
Beikmohammadi & Faez (2018)	MobileNet architecture	Logistic regression classifier	LeafSnap	90.54
Cıbuk et al. (2019)	AlexNet & VGG16 architectures	SVM	Flower17	96.39
Cıbuk et al. (2019)	AlexNet & VGG16 architectures	SVM	Flower102	95.70
Pawara et al. (2017)	AlexNet & GoogLeNet architectures	CNN (Fine-tuning)	Swedish	99.92
Pawara et al. (2017)	AlexNet & GoogLeNet architectures	CNN (Fine-tuning)	Folio	98.60
Zhang et al. (2015)	7-layer CNN architecture		Flavia	94.60
Wick & Puppe (2017)	9-layer CNN architecture	Flavia	99.81
Wick & Puppe (2017)	9-layer CNN architecture		Foliage	99.40
Barré et al. (2017b)	17-layer CNN architecture	Flavia	97.90
			Foliage	95.60
			LeafSnap	86.30
Lee et al. (2017)	AlexNet architecture	Multilayer Perceptron Classifier (MLP)	Flavia	99.50
Lee et al. (2017)	AlexNet architecture	Multilayer Perceptron Classifier (MLP)	Folio	99.40
Sulc & Matas (2015)	ResNet152 & Inception-ResNetv2 architectures based on LBP	CNN (Fine-tuning)	Flavia	99.80
			Foliage	99.30
			Swedish	99.80
			LeafSnap	83.70
Kaya et al. (2019)	AlexNet & VGG16 architectures	CNN (Fine-tuning) & LDA	Flavia	99.10
Kaya et al. (2019)	AlexNet & VGG16 architectures	CNN (Fine-tuning) & LDA	Swedish	99.11
Xiao et al. (2018)	Inceptionv3 with Attention Cropping		Flower102	95.10