Medicinal plants have always been essential to preserving human health, particularly in areas where access to contemporary healthcare is scarce, as almost 80% of the population uses plant-based medicines, according to the World Health Organization. This research introduces SegUNet, a revolutionary deep learning technology that blends SegNet and UNet architectures to detect diseases in Tulsi plants. Tulsi poses special difficulties for disease identification because of its unusual characteristics, which include tiny leaves, reticulate leaf venation, and serrated margins. We train two semantic segmentation models and combine them to create the hybrid SegUNet model, which attains a remarkable 98.61% accuracy and a weighted precision of 94.87%. This study, which departs from conventional techniques, opens the door for a web-based solution for Tulsi disease detection, promising transformative effects on agricultural diagnostics and disease management.