[36] | 2021 | Multi-task | Corn/maïze | leaf | Gray leaf spot | 5 severity levels | Self-collected | proposed CNN | Aaccuracy of 95.33% |
[19] | 2021 | Single | Wheat | Spike/ fruit | Fusarium head blight | 0, 1, 2, 3, 4 and 5 | Self-collected (3.600 images) | AlexNet and Random Forest | * |
[7] | 2022 | single | Apple | leaf | Alternaria Leaf Blotch | healthy (0), early (0 - 0.95%), mild (0.95% - 1.75%), moderate (1.75% - 3.00%) and severe (3.00% - 100%) | Self-collected dataset of 5382 samples | DeeplabV3+, Unet PSPNet, VGG, ResNet and MobileNetV2 | Mean accuracy of 96.41%. |
[45] | 2022 | Single | Strawberry | Leaf | Gray mold | percentage | Self-collected dataset of 400 samples | Unet, XGBoost, K-means, Otsu | IoU accuracy, pixel accuracy, and dice accuracy are 82.12%, 98.24% and 89.71% respectively. |
[46] | 2022 | Single | ¨Paddy | Leaf | Bacterial blight | healthy, infected but disease is not severe, and infected and disease is severe | Self-collected (650 samples) | proposed-CNN | Accuracy of 97.692% |
[47] | 2022 | Single | Patato | leaf | potato blight | 1% - 20%, 21% - 40%, 41% - 60%, 61% - 80%, and 81% - 100% | Self-collected dataset (9600 images) | proposed CNN | Accuracy: 86.625% |
[39] | 2022 | Single | Tomato | leaf | Begomovirus | 4 severity levels | Self-collected | proposed CNN | * |
[48] | 2022 | Single | Tomato | leaf | * | percentage | Internet | MRNN | * |
[16] | 2022 | Single | Paprika | leaf & fruit | Blossom end rot, ray mold, powdery mildew, snails and slugs, spider mite, and Cercospora | 11 severity levels. | Self-collected (6.000 images) | proposed-CNN | Mean average Precision: 91.7% for the abnormality detection; Mean panoptic quality score: 70.78% for severity level prediction. |
[5] | 2022 | Single | Maize | leaf | Maydis leaf blight | low, medium and high | Self-created (1.760 images) | proposed-CNN, VGG16, VGG19, ResNet50, InceptionV3, Xception, DenseNet121, MobileNetV2 and NASNetMobile | Proposed-CNN (accuracy: 99.13% and f1_score: 98.97%) |