<p>Medicinal plants play a vital role in traditional medicine, modern pharmaceuticals, and the food industry. Given their significance for human health and diverse industrial applications, rapid and accurate identification using machine vision techniques has become essential. However, the high visual similarity among different species and the complexity of image data poses significant challenges to automatic classification. This study proposes an innovative hybrid model named QGCVT, which integrates convolutional, transformer, and quantum-inspired graph structures. The model operates as follows: First, the input image is divided into small patches, and structural features are extracted using Convolutional Vision Transformer (CvT) blocks. Next, a dynamic graph is constructed from these patches, and a quantum-inspired propagation mechanism drawing on the principles of superposition and entanglement is applied to capture complex relationships between different image regions. Finally, enhanced feature representations produced by quantum-inspired Graph Convolutional Network (GCN) layers are used for accurate classification of medicinal plant species. The proposed QGCVT model was evaluated on four real-world datasets containing species with high visual similarity, varying lighting conditions, and diverse backgrounds. Experimental results demonstrate that the model achieved accuracies of 99.24%, 99.47%, 99.98%, and 100% on the four datasets, respectively, with the perfect score indicating error-free classification on the final dataset. These results highlight QGCVT’s exceptional capability in extracting and integrating local and global features from complex images, delivering state-of-the-art performance in medicinal plant species recognition. Consequently, QGCVT represents a powerful tool for automated medicinal plant identification, smart agriculture, and biodiversity conservation efforts.</p>