Table 13.

Comparison of various reviewed papers.

S. No.	Year and Reference	Species	Techniques Used	Disease Identified	Performance Measure	Value
1.	2010 [2]	Sugar beet	SVM based on hyperspectral reflectance	Sugar beet rust, Cercospora leaf spot, powdery mildew	Accuracy	Higher than 86%
2.	2012 [35]	Grapes, Wheat	Backpropagation networks, image processing technologies	Grape (downy mildew, powdery mildew), wheat (stripe rust, leaf rust)	Accuracy (prediction accuracy, fitting accuracy)	Fitting accuracy—100% (for both), prediction accuracy—97.14% (grape), 100% (wheat)
3.	2012 [3]	Apple	Image processing techniques (multi-class SVM)	Apple rot, apple scab, apple blotch	Accuracy	93%
4.	2013 [72]	Sugar beet	Spectral disease indices	Sugar beet rust, cercospora leaf spot, powdery mildew	Accuracy	Sugar beet rust—87%, cercospora leaf spot—92%, powdery mildew—85%
5.	2013 [37]	Grape	Feed-forward back propagation neural network	Powdery mildew, downy mildew	Accuracy	100% (using the HUE feature only)
6.	2016 [38]	-	SVM (serial combination of two SVMs)	Thrips, Tuta absoluta, leaf miners (damaged by pest insects), early blight, powdery mildew, late blight (pathogens symptoms)	Accuracy	87.80%
7.	2016 [9]	Cucumber	Convolutional neural network	KGMMV, WMV, PRSV, CMV, CCYV, ZYMV, MYSV	Accuracy	82.3%
8.	2017 [41]	Tomato	Deep learning (AlexNet and SqueezeNet)	Spider mites, yellow leaf curl virus, early blight, bacterial spot, septoria leaf spot, leaf mold, late blight, mosaic virus, target spot	Accuracy	AlexNet—95.65%, SqueezeNet—94.3%
9.	2017 [10]	Tomato	CNN	Yellow leaf curl virus, bacterial spot, late blight, leaf mold, spider mites, septoria spot, mosaic virus, target spot, early blight	Accuracy	99.18%
10.	2017 [42]	Apple	Deep convolutional neural network (AlexNet)	Rust, mosaic, alternaria leaf spot, brown spot	Accuracy	97.62%
11.	2018 [61]	25 different plant species	CNN models based on deep learning techniques	58 distinct classes	Accuracy	99.53%
12.	2018 [4]	Papaya	Random forest (RF)	Healthy/unhealthy	Accuracy	70%
13.	2018 [13]	Cucumber	DCNN	Downy mildew, anthracnose, powdery mildew, and target leaf spots	Accuracy	93.4%
14.	2018 [14]	Tomato	CNN with learning vector quantization	Septoria spot, bacterial spot, yellow curved, and late blight	Accuracy	86%
15.	2018 [5]	Orange	SVM with K-means clustering (classification), degree of disease severity—fuzzy logic	Brown rot, citrus canker, melanoses, stubborn	Accuracy	90%
16.	2019 [16]	13 different plant leaves (grape, apple, tomato, cherry, peach, potato, and others)	Nine-layer deep CNN	Potato (early blight), cherry (powdery mildew), apple with black rot, peach with bacterial spots, tomato (leaf mold), grape (leaf blight), etc.	Accuracy	96.46%
17.	2019 [62]	Apple, tomato	Convolutional neural network	Healthy/diseased	Accuracy	87%
18.	2020 [20]	Grape	Convolutional neural network (UnitedModel)	Esca, black rot, isariopsis	Validation accuracy, test accuracy	test accuracy—98.57%, validation accuracy—99.17%
19.	2019 [70]	Cotton, tomato	Image processing techniques, neural network	Cotton (target spot, bacterial leaf spot), tomato (septoria leaf spot, leaf mold)	Accuracy	For cotton (bacterial leaf spot—90%, target spot)—80%, for tomato (septoria leaf spot and leaf mold)—100%
20.	2019 [17]	Cucumber	CNN with global average pooling	Black spot, powdery mildew, angular leaf spot, gray mold, anthracnose, downy mildew	Accuracy	94.65%
21.	2019 [7]	Chili	SVM	Cucumber mosaic virus	Accuracy	57.1%
22.	2019 [64]	Guava	Deep convolutional neural network	Rust, algal leaf spot, whitefly	Accuracy	98.74%
23.	2019 [46]	Guava	Convolutional neural network	Anthracnose, fruit canker, fruit rot	Accuracy	95.61%
24.	2019 [47]	Lady finger	SVM, artificial neural network	Powdery mildew, leaf spot, yellow mosaic vein	Accuracy	85% (SVM) and 97% (ANN); without noise, 92% (SVM) and 98% (ANN)
25.	2019 [19]	Pearl millet	Transfer learning with feature extraction	Mildew	Accuracy, f1-score, recall, precision	Accuracy—95%, f1-score—91.75%, recall—94.50%, precision—90.50%
26.	2021 [24]	Soybean	CNN (GoogleNet, AlexNet)	Brown spot, frogeye leaf spot, bacterial blight	Accuracy	98.75% (AlexNet), 96.25% (GoogleNet)
27.	2020 [68]	Tomato	Convolutional neural network	Septoria leaf spot, early blight, mosaic virus, yellow leaf curl virus, bacterial spot	Accuracy	97%
28.	2021 [26]	14 crops	Discount momentum deep learning optimizer	26 disease classes	Accuracy	97%
29.	2020 [51]	Mango	Feed-forward neural network (deep neural networks)	Powdery mildew, gall midge, anthracnose	Accuracy	91.32% (training accuracy), 85.45% (testing accuracy)
30.	2020 [22]	Potato, tomato, bell pepper	CNN	Potato (early and late blight), bell pepper bacterial spot, tomato (target spot, mosaic virus, early blight, bacterial spot, yellow leaf curl virus, late blight, septoria leaf spot, spider mites, leaf mold)	Test Accuracy	88.8%
31.	2021 [73]	Peach	Hybrid approach (convolutional autoencoder, convolutional neural network)	Bacterial spot	Accuracy	Testing accuracy—98.38%, training accuracy—99.35%
32.	2022 [56]	14 crops	Deep ensemble neural network	38 classes	Accuracy	99.99%
33.	2021 [28]	Tomato	C-GAN (for producing synthetic images), DenseNet	Two-spotted spider mite, bacterial spot, septoria leaf spot, yellow leaf curl virus, target spot, early blight, leaf mold, late blight, mosaic virus	Accuracy	99.51% (5 classes), 98.65% (7 classes), 97.11% (10 classes)
34.	2021 [71]	26 plant species	LFM-CNAPS based on meta-learning	60 diseases	Accuracy	93.9%
35.	2021 [31]	Grape	CNN (VGG, DenseNet, ResNet)	Black rot, leaf blight, esca	Accuracy	98.27% (DenseNet accuracy)
36.	2021 [32]	Tomato	GoogleNet, VGG16	Bacterial spot, early blight, late blight	Accuracy	GoogleNet—99.23%, VGG16—98%
37.	2021 [1]	Apple	Convolutional neural networks	Bitter rot, powdery mildew, sooty blotch	Accuracy	97%
38.	2022 [66]	Tomato	Optimal mobile network-Based CNN	Late blight, target spot, leaf mold, and early blight	Accuracy, recall, precision, kappa, F-score	98.7% (accuracy), 0.9892 (recall), 0.985 (precision, F1-score, kappa)