Hazelnut42/Fruit_Detection

🍎 Fruit Ninja: Multi-Class Fruit Detection

Detect 10 types of fruits using YOLOv8 deep learning model!

🎯 Demo

Upload an image containing fruits, and the model will detect and classify them automatically.

Supported Fruits:

• —🍎 Apple
• —🥑 Avocado
• —🫐 Blueberry
• —🫑 Capsicum
• —🍒 Cherry
• —🥝 Kiwi
• —🥭 Mango
• —🍊 Orange
• —🍈 Rockmelon
• —🍓 Strawberry

📊 Model Performance

Per-Class Performance (AP@0.5)

🔧 Model Details

• —Architecture: YOLOv8n (Nano)
• —Parameters: 3.0M
• —Dataset: deepNIR (922 images, 10 classes)
• —Training: 100 epochs, 27.8 minutes on Tesla T4
• —Framework: Ultralytics YOLOv8

🚀 Training Configuration

TRAINING_PARAMS = {
    'epochs': 100,
    'batch': 16,
    'imgsz': 640,
    'lr0': 0.01,
    'optimizer': 'AdamW',

    # Data Augmentation
    'mosaic': 1.0,
    'mixup': 0.1,
    'hsv_h': 0.015,
    'hsv_s': 0.7,
    'hsv_v': 0.4,
    'degrees': 10.0,
    'translate': 0.1,
    'scale': 0.5,
    'fliplr': 0.5,
}

📝 How to Use

1. 1.Upload an image containing fruits
2. 2.Adjust confidence threshold (optional)
3. 3.Click "Detect Fruits"
4. 4.View results with bounding boxes and labels

🎓 Academic Context

This model was developed as part of a Computer Vision course homework assignment (HW3: Fruit Ninja). The goal was to train a deep learning model for multi-class fruit detection suitable for agricultural automation applications.

Real-world Applications:

• —Automated fruit counting in orchards
• —Quality control in packing facilities
• —Yield estimation for harvest planning
• —Robotic fruit picking systems

📚 References

• —Ultralytics YOLOv8
• —YOLOv8 Documentation
• —deepNIR Dataset

📄 License

MIT License - Feel free to use for educational purposes!

Developed by: [Your Name] Course: Computer Vision Date: October 2025