Plant health monitoring and pest handling are common practices in the agricultural domain, and the outcome achieved with this process helps to get complete information about the infection. This research considered the palm-tree image data for the examination, and it helps to detect the healthy/disease class of the date palm tree with improved results. This work considered the digital images collected with a thermal camera for examination and to achieve a better outcome, it considered the traditional DL models like VGG16, ResNet models, and DenseNet models for the examination. The different phases of this research include the following: data collection and resizing it to 224x224 pixels, feature extraction with a DL model, feature reduction, and generation of fused-features (FF) vector, detection using machine- learning classifiers, and confirming the result with 3-fold cross-validation. This work presented individual- features and fused-features based detection. This work considered the SoftMax and other related classifiers for examination. The outcome of this study confirms that the developed scheme can help to achieve a detection accuracy of >96% with the thermal image database. This work considered 700 images per class after augmentation, and the proposed tool works on this thermal image database and helps to achieve better results.
Manic et al. (Wed,) studied this question.