Object Detection Using YOLOv8

This project delivers a comprehensive solution for license plate detection using the state-of-the-art YOLOv8 object detection model. From dataset preparation and model training to developing a robust Flask API, this repository is your one-stop guide to implementing real-time license plate detection.

---

🚀 Project Highlights

• Oriented Bounding Boxes (OBB): Handles rotated license plates with precision.
• Flask API Integration: Provides an easy-to-use web interface for detection.
• GPU-Accelerated Training: Powered by Kaggle for efficient model training.
• Visualization-Ready: Clear and insightful results showcased in a Jupyter Notebook.

---

🛠️ Features

1️⃣ Dataset Preparation

• Dataset sourced from Roboflow with high-quality annotations.
• Preprocessing steps include scaling, augmentations, and proper formatting for YOLOv8.

2️⃣ YOLOv8 Training

• Trained on a Kaggle GPU environment for optimal performance.
• Model trained for 10 epochs (for better prediction we can train for more epochs).
• The best model weights (best.pt) are ready for deployment.

3️⃣ Flask API

• User-Friendly Interface: Upload images via the web interface for detection.
• AI-Powered Backend: Returns:
  • Images with annotated bounding boxes.
  • JSON files with bounding box coordinates and class labels.

4️⃣ Inference and Visualization

• Intuitive visualization of results through bounding boxes and JSON outputs.
• Detection results include bounding box coordinates, angles, and class labels.

---

📊 Example Results

See the YOLOv8 model in action below:

Example 1:
Detected rotated license plate with oriented bounding boxes.

Example 2:
License plate detected with high accuracy.

Example 3:
Detection of multiple plates within a single image.

Example 4:
Another example of a detected license plate with accurate bounding box positioning.

---

⚙️ Project Workflow

1. Dataset Preparation

   • Downloaded and preprocessed the dataset from Roboflow.

2. Model Training

   • Trained the YOLOv8 model in a Kaggle GPU environment.
   • Saved the trained model weights as best.pt.

3. Inference

   • Performed inference using the trained model on test images.
   • Saved results as annotated images and bounding box coordinates and rotated angle details.

4. Flask API

   • Built APIs to handle image uploads and run the YOLOv8 model for real-time inference.

---

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/Ankitjha2202/object_detection_services.git
cd object_detection_services