Procedia Computer Science · 2024 · 12 citations · 22 references
ComputingComputer SciencePlant PathologyComputer VisionMachine LearningPlant Disease DiagnosisClassificationDeep LearningDisease DetectionFeature-level FusionPattern RecognitionSupport Vector MachinesConvolutional Neural NetworksDeep Feature FusionDeep Feature LearningDeep Learning-based ClassificationDeep Learning-based DetectionDeep Learning-based DiagnosisDeep Learning-based FusionDeep Learning-based Identification
In order to ensure grape output and quality, it is necessary to identify grape leaf diseases quickly and accurately. By merging deep features from convolutional neural networks (CNNs) and support vector machines (SVMs), we provide an efficient identification approach. To improve the accuracy of illness identification, we investigate several CNN designs, deep feature layers, and fusion methods. Deep features are retrieved using certain CNN layers after scaling images to match CNN input criteria. The performance of classification is improved by combining two different kinds of deep features from various networks. An SVM classifier for several classes is trained using the combined deep features. A public dataset experiment's findings show that combining deep features enhances classification performance compared to utilizing a single type of deep value. The best classification score is achieved by directly concatenating particular deep features from ResNet50 and ResNet101, which results in an astounding F1 score of 99.82%. Furthermore, the SVM classifier created using this fusion approach exhibits classification performance equivalent to training a CNN model, but with a training period that is noticeably shorter—less than one second. With the help of this method, grape leaf diseases may be quickly and precisely identified, satisfying the demands of actual agricultural operations.
22
Plant leaf disease classification using EfficientNet deep learning model
Ümit Atila, Murat Uçar, Kemal Akyol et al. · Ecological Informatics · 2020 · 897 citations
Attention embedded residual CNN for disease detection in tomato leaves
R. Karthik, M. Hariharan, Sundar Anand et al. · Applied Soft Computing · 2019 · 557 citations
Recent Advances in Crop Disease Detection Using UAV and Deep Learning Techniques
Tej Bahadur Shahi, Cheng‐Yuan Xu, Arjun Neupane et al. · Remote Sensing · 2023 · 255 citations · Full text