# Open Agri - Plant Disease Detection AI

This project focuses on identifying plant diseases using advanced Deep Learning models. It includes a user-friendly web interface powered by a multimodal Large Language Model (LLM) and specialized pipelines for Tomato Disease detection.

📂 Project Structure

• new/: Contains the interactive web application.
  • main.py: A Gradio-based interface utilizing the YuchengShi/LLaVA-v1.5-7B-Plant-Leaf-Diseases-Detection model for analyzing leaf images, describing symptoms, and suggesting treatments.
• tomato-disease-ai/: A comprehensive pipeline for tomato disease classification and segmentation.
  • dataset/ & segmented_dataset/: Training and validation data directories.
  • models/: Storage for trained Keras/H5 models.
  • src/: Training scripts (train.py), utilities (utils.py), and evaluation scripts.
  • segmentation/ & patch_classifier/: Specialized scripts for image segmentation and patch-based classification.

🚀 Getting Started

Prerequisites

Ensure you have Python installed. It is recommended to use a virtual environment or Conda environment to manage dependencies.

Installation

1. Clone the repository:

   git clone https://github.com/Prathu241/open_agri.git
   cd open_agri

2. Install General Dependencies (for tomato-disease-ai):

   cd tomato-disease-ai
   pip install -r requirements.txt

3. Install Dependencies for the Web App (new/): The web app requires additional libraries like gradio, torch, and transformers.

   pip install gradio torch transformers opencv-python pillow accelerator

🎮 Usage

1. Run the Web Application

Option A: Streamlit (Recommended)

This provides a polished, interactive web UI.

cd new
streamlit run streamlit_app.py

Option B: Gradio (Legacy)

cd new
python main.py

This will launch a local Gradio server (usually at http://127.0.0.1:7860). Upload an image to get a detailed disease analysis.

2. Train the Tomato Disease Model

To train the custom classifiers or segmentation models:

cd tomato-disease-ai/src
python train.py

(Refer to individual scripts in tomato-disease-ai for specific segmentation or patch-based training tasks.)

🧠 Models Used

• LLaVA-v1.5-7B-Plant-Leaf-Diseases-Detection: A fine-tuned multimodal model for comprehensive plant disease diagnosis.
• Custom CNNs/SegFormer: Tailored models for high-accuracy tomato leaf disease segmentation and classification.