Integrating Deep Learning and GIS for Herbal Plant Identification and Location Finding System

This research explores the utilization of deep learning algorithms for the identification of herbal plants to address the difficulties arising from the numerous variations and similarities among plant species. The conventional identification of plants is mostly time-consuming, laborious, and prone to human error, requiring much expertise. This study demonstrates how the CNN model can be generated and integrated into a web-based and mobile application for plant recognition. The model was trained on a dataset of 800 sample images and achieved a training accuracy rate of 99.2%, suggesting its ability to classify herbal plants correctly using the training data. The model was also tested with 500 samples, comprising 50 testing images per class, and achieved an overall accuracy rate of 85.2%. The study demonstrates the ability of the CNN model to classify herbal plants. It also further exhibits the system's potential in pinpointing the locations of identified species for allowing the users to find invaluable resources for locating herbs in the area through GIS maps. This approach simplifies the identification and location detection of herbal plants for a wide range of users, ranging from researchers and medical professionals to non-experts and other stakeholders interested in studying or using herbal plants in the medical field.