Wheat Detection with YOLO 🚜🌾

Welcome to the Wheat Detection project! This repository uses the YOLO11 model to detect wheat heads in agricultural images, leveraging the GlobalWheat2020 dataset. Perfect for precision agriculture and crop monitoring! 🌱

📖 Overview

This project implements wheat head detection using the Ultralytics YOLO framework. It includes downloading the GlobalWheat2020 dataset, training a YOLO11n model, and performing inference on uploaded images. The results are visualized with bounding boxes around detected wheat heads. 🌾

🚀 Features

• 📥 Downloads and organizes the GlobalWheat2020 dataset
• 🛠️ Trains a YOLO11n model for wheat head detection
• 🖼️ Performs inference on uploaded images
• 📊 Visualizes results with bounding boxes using Matplotlib
• ⚙️ Easy-to-use Python script for end-to-end workflow

📋 Requirements

• Python 3.8+
• Ultralytics YOLO (pip install ultralytics)
• Matplotlib
• NumPy
• Pillow
• Google Colab (optional for running in the cloud ☁️)

Install dependencies:

pip install ultralytics matplotlib numpy pillow

🛠️ Installation

1. Clone the repository:

   git clone https://github.com/shervinnd/wheat-detection.git
   cd wheat-detection

2. Install required packages:

   pip install -r requirements.txt

3. Run the script in a Python environment or Google Colab:

   python wheat_detection.py

📂 Dataset

The project uses the GlobalWheat2020 dataset:

• Images and annotations are downloaded from Zenodo and Ultralytics.
• The dataset is organized into images, labels, and annotations directories.
• A YAML file (GlobalWheat2020_subset.yaml) is created for training and validation subsets.

🧠 Model Training

The script trains a pre-trained YOLO11n model on the GlobalWheat2020 dataset for 5 epochs. Adjust the number of epochs or image size as needed:

model.train(data="GlobalWheat2020_subset.yaml", epochs=5, imgsz=640)

🔍 Inference

Upload an image to perform wheat head detection:

1. The script prompts for image upload.
2. Results are displayed with bounding boxes using Matplotlib.

Example:

results = model("path/to/image.jpg")
for result in results:
    img = result.plot()
    plt.imshow(img)
    plt.show()

📊 Results

The trained model outputs bounding boxes around wheat heads in images. Results are saved in the runs/detect directory.

Contributions are welcome! 🌟 Feel free to:

• Open issues for bugs or feature requests
• Submit pull requests with improvements
• Add new datasets or enhance the model

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.

📬 Contact

For questions or feedback, reach out via GitHub Issues or email at shervindanesh8282@gmail.com

Happy wheat detecting! 🌾🚀