Bio-optimized complex valued spatiotemporal GNN for herbal species classification

Herbs have long been integral to various health and medicinal benefits. Identifying the correct herb species from thousands of diverse options is a laborious and time-consuming process. To address this issue, an automated computer vision system that reduces the traditional labor-intensive work of herbal species classification is needed. In this paper, we propose a novel method for automated herb classification via a complex-valued spatiotemporal graph convolutional neural network (AHC-CVSTGCN) with hierarchical manta ray foraging optimization. The objective of our research work is to develop a robust, effective, and optimized classification model driven by graph neural networks on a diverse array of morphological features of various herbal leaves. Input images are collected from FLAVIA and the medical leaf dataset. The images are first preprocessed utilizing Multiple Local Particle Filter (MLPF) to remove background noise and enhance quality, and then Revised Tunable Q-Factor Wavelet Transform (RQFWT) extracts relevant features such as shape, color, and texture. Finally, CVSTGCN classifies the images, with HMRFO optimization applied to further improve accuracy. Our proposed approach bridges the gaps in the classification of various herbal species, empowering medicine practitioners to make informed decisions. Experimental evaluation demonstrated that our approach significantly outperforms existing methods by achieving 99.40% high accuracy, 99.11% high precision, and 99.12% high recall. By contributing a reliable and effective solution for automated herbal species classification, this work presents a crucial paradigm shift for medicinal plant science and health care practitioners.