Table 13.
A comparative analysis of studies on potato disease detection.
| Author | Disease type | Algorithm | Dataset | Result |
|---|---|---|---|---|
| Oppenheim et al. [30] | Black scurf, silver scurf, common scab, and black dot | Custom CNN model | Custom prepared 2,465 images | 96% accuracy |
|
| ||||
| Oishi et al. [36] | Not mentioned | Fast R-CNN and YOLO v3 | Pascal VOC 2007, COCO dataset, PlantVillage, and custom prepared | Fast R-CNN with 96.7% accuracy |
|
| ||||
| Afonso et al. [37] | Blackleg | ResNet18 and ResNet50 | Custom prepared (532 images) | 91% recall |
|
| ||||
| Hou et al. [38] | General degree and a serious degree of both early blight and late blight | k-NN, SVM, RF, and ANN | AI challenger global AI contest (2840 images) | SVM with 97.4% accuracy |
|
| ||||
| Tiwari and Divyansh [39] | Early blight and late blight | VGG16, VGG19, Inception-v3, and LR | PlantVillage dataset (2,152) | VGG19 with LR 97.8% accuracy, 97.8% precision, 97.8% recall, and 97.8% F1-score |
|
| ||||
| Gao et al. [41] | Late blight | SegNet | Custom prepared 2,100 images | Not mentioned |
|
| ||||
| Lee et al. [43] | Early blight | Proposed model using CNN, VGG16, and VGG19 | Not mentioned | The proposed model scored 99% accuracy |
|
| ||||
| Iqbal and Talukder [44] | Early blight and late blight | RF, LR, k-NN, DT, NB, LDA, and SVM | Custom prepared 450 images | RF scored 97% accuracy |
|
| ||||
| Asif et al. [46] | Early blight and late blight | AlexNet, VggNet, ResNet, LeNet, and sequential model | Kaggle, dataquest dataset, and custom prepared dataset | The proposed CNN model scored 97% accuracy |
|
| ||||
| Patil et al. [48] | Early blight and late blight | SVM, RF, and ANN | Custom prepared 892 images and PlantVillage (300 images) | ANN scored 92% accuracy |
| Sholihati et al. [53] | Alternaria solani, phytophthora infestans, virus, and insect | VGG16, proposed model, and VGG19 | 5,200 open-source datasets | Proposed model 91% accuracy, 88% precision, and 89% recall |
|
| ||||
| Tarik et al. [54] | Roll virus, hollow heart, scab, soft rot, sutali poka rrog, virus jonito rog, and early blight | Custom-built CNN | Custom prepared 2034 images | 99.23% accuracy |
|
| ||||
| Sert [92] | Early blight and late light | Faster R-CNN and GoogLeNet, SequezeNet, and AlexNet | Plant village and custom prepared dataset | Faster R-CNN with GoogLeNet scored 98.06% accuracy, 98% precision, 98% recall, and 98% F1-score |
|
| ||||
| Rashid et al. [97] | Early blight and late blight | Custom-built CNN | Custom prepared 4062 images | 99.75% accuracy, 99.6% precision, 99.6% recall, and 99.6% F1-score |