Answers are grounded in this paper and cite their sources.
The medicinal plants are complex in their phytochemistry and morphology that affects the quality of medicinal efficacy. The process of automated evaluation of leaf quality needs powerful feature extraction and smart algorithms that can distinguish between normal and diseased or stressed leaves. This paper puts forth a multimodal model to choose predictive characteristics to evaluate the quality of medicinal plants. The Ocimum leaf (Tulsi) dataset was collected, the hierarchical structure of the data was validated, and the images were enhanced using bilateral filtering to remove noise while preserving the venation. Morphological (area, perimeter, aspect ratio and circularity) texture histograms are derived and normalized Local Binary Pattern (LBP) texture histograms. These handcrafted features are complemented by a lightweight MobileNetV2 deep branch. The most informative features are then chosen using statistical analysis and mutual information ranking to be used in the classification. A multimodal fusion net is trained to fuse the chosen features and the deep representations. Comparative evaluations reveal that the proposed approach displays enhanced accuracy, precision, recall, and F1-score in contrast to traditional models like DenseNet121 and VGG. The findings indicate that combining morphological and texture features with deep features enhances resistance to changes in lighting and deformation of leaves. The study offers a computationally efficient approach to the quality assessment of medicinal plants and the basis of interpretable monitoring devices in the manufacturing of herbal medicines.
No takes yet. Share an insight, caveat, or question.
Lambture et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: