Fruit Detection with YOLOv8

A machine learning project that detects and classifies 9 types of fruits using YOLOv8 object detection.

Fruits Detected

• Apple
• Banana
• Cherry
• Chickoo
• Grapes
• Kiwi
• Mango
• Orange
• Strawberry

Dataset

• Total images: 332
• Annotated for training: 64 images
• Tool used: Roboflow

Model

• Architecture: YOLOv8 Nano
• Framework: Ultralytics
• Training device: Apple M3 (MPS)
• Epochs: 50

Project Structure

fruit_detection_project/
├── scripts/
│   ├── 1_explore_dataset.py
│   ├── 2_select_images.py
│   ├── 4_train_model.py
│   └── test_custom_image.py
└── runs/detect/fruit_detector/
    └── weights/best.pt

How to Use

Train the Model

python3 scripts/4_train_model.py

Test on Custom Image

python3 scripts/test_custom_image.py

Requirements

pip install ultralytics opencv-python

Results

Model trained on 64 annotated images with transfer learning from YOLOv8n pretrained weights.