https://github.com/ankitjha2202/object_detection_services
License plate detection using yolov8 oriented bounding boxes
https://github.com/ankitjha2202/object_detection_services
computer-vision flask rest-api yolov8 yolov8-detection
Last synced: about 1 month ago
JSON representation
License plate detection using yolov8 oriented bounding boxes
- Host: GitHub
- URL: https://github.com/ankitjha2202/object_detection_services
- Owner: Ankitjha2202
- License: other
- Created: 2024-11-15T13:57:26.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-11-18T04:41:19.000Z (over 1 year ago)
- Last Synced: 2025-03-16T09:29:38.846Z (over 1 year ago)
- Topics: computer-vision, flask, rest-api, yolov8, yolov8-detection
- Language: Jupyter Notebook
- Homepage:
- Size: 70.7 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: license-plate-detection.ipynb
Awesome Lists containing this project
README
# 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
```bash
git clone https://github.com/Ankitjha2202/object_detection_services.git
cd object_detection_services