ANPR: Automatic Number Plate Recognition using YOLOv8

A deep learning-based ANPR system built and trained from scratch on a custom Indian road dataset for robust number plate detection using YOLOv8.

---

🚀 Highlights

• 📸 Custom Dataset: 700+ images captured on Indian roads with varying lighting, occlusion, and angles.
• 🏧 Annotation: Annotated using LabelImg in YOLO format.
• 🧠 Model: YOLOv8 trained from scratch.
• 📊 High Performance: Strong evaluation metrics on the test set.

---

📂 Project Structure

ANPR/
├── dataset/
│   ├── images/
│   ├── labels/
│   └── ...
├── runs/
│   └── detect/
├── results.csv            # Evaluation metrics
├── yolov8_train.py        # Training script
├── inference.py           # Inference pipeline
├── requirements.txt
└── README.md

---

🗾 Dataset

• Total Images: 700+
• Annotations: YOLO format via LabelImg
• Conditions: Includes diverse road scenarios from Indian streets

---

🏋️‍♂️ Training

• Model: YOLOv8n / YOLOv8s
• Epochs: 100
• Image Size: 640x640
• Framework: Ultralytics YOLOv8

yolo task=detect mode=train model=yolov8n.yaml data=data.yaml epochs=100 imgsz=640

---

📈 Evaluation

Evaluation metrics from results.csv:

Metric	Value
mAP@0.5	0.957
mAP@0.5:0.95	0.780
Precision	0.963
Recall	0.947
F1 Score	0.955

Evaluated on a separate test set.

---

🧪 Inference

Run inference on an image:

yolo task=detect mode=predict model=best.pt source='path/to/image.jpg'

🔍 Output Preview

Detected Plates

---

🔧 Requirements

pip install -r requirements.txt

• Python 3.8+
• OpenCV
• Ultralytics
• PyTorch

---

📌 TODO

• Add OCR for license plate text recognition
• Streamlit Web UI
• Real-time video tracking support

---

📃 License

Licensed under the MIT License.

---

🙌 Acknowledgements

• Ultralytics YOLOv8
• LabelImg