Automatic Identification of Medicinal Plants with Transfer Learning

Medicinal plants are widely recognized for their medicinal properties, so accurate identification is very important for scientists, pharmacists and people. This work presents a transfer learning-based solution to identifying medicinal plants using MobileNetV2 architecture. Transfer learning can effectively classify plant species using pre- trained models through optimization of plant data. The system is designed to take user input images, process images from the MobileN etV2 model, and predict vegetation with high accuracy. After categorization, the system integrates the Google Custom Search API to provide useful information to users, including the top three web sites describing the use of medicinal herbs. This article describes the architecture, processes, and procedures involved in developing Web applications. Our method simplifies the plant identification process using adaptive learning, thereby reducing training time and computational resources while increasing classification accuracy. The solution is designed to support veterinarians, teachers, and assistants in quickly identifying medicinal plants and understanding their uses, thus improving the overall understanding of phytotherapy.