Project Overview
Traditional medicinal plant identification requires expert knowledge or extensive manuals. This project automates the identification process using deep learning, making medicinal plant recognition accessible and instant.
The system processes plant images through a convolutional neural network trained on the Mendeley Indian Medicinal dataset, covering over 1,500 images across 30 species with 98% accuracy.
Technical Architecture
Frontend: User-friendly web interface for image upload and result display with real-time processing feedback.
Backend: Flask application hosted on Google Cloud, handling image preprocessing, model inference, and result formatting through RESTful APIs.
Model: Fine-tuned ResNet architecture optimized for medicinal plant classification with data augmentation and transfer learning techniques.