ObjectDetection

This project provides a complete pipeline for vehicle and license plate detection and tracking. It uses a YOLO model trained specifically for license plate detection and another YOLO model for car detection. The SORT algorithm is integrated to track detected cars across video frames. Detected license plates are associated with their corresponding vehicles, and Optical Character Recognition (OCR) is applied to read the license plate numbers from the detected regions.

Example

Here is an example :

Instructions

1. Environment Setup

• Install the required dependencies using the requirements.txt file in the ObjectDetection directory:

  pip install -r requirements.txt

2. Data Preparation

• Clone the OIDv4_ToolKit repository inside a software folder within ObjectDetection:

  mkdir software
  cd software
  git clone https://github.com/EscVM/OIDv4_ToolKit.git

• Use OIDv4_ToolKit to extract your dataset (e.g., labeled license plate images). The extracted data will be in the Dataset/ directory, containing train, test, and validation folders.

• If using license plate data, you will find a folder named "Vehicle registration plate". Group all images into a new images folder inside the "Vehicle registration plate" folder, rename the Labels folder to Labels_Not_Yolo, and create an empty labels folder.

3. Label Conversion

• Use the scripts in yolo/convert_labels_to_yolo_format to convert your labels to YOLO format.
• You can verify the label conversion using the verify_picture_label script.

4. Model Training and Testing

• To train the model, use the train_license_plate_model script.
• To test the model, use the test_license_plate_model script.

Note: Make sure you have the appropriate dataset prepared before training or testing.

5. Tracking Setup

• Before using main.py and util.py, clone the SORT repository inside the yolo directory:

  cd yolo
  git clone https://github.com/abewley/sort.git

• Create a video folder inside yolo and place your video file (e.g., car.mp4) there.

6. Running the Project

• Follow the scripts and instructions in the respective folders to process your data, train, test, and run tracking on your videos.

For any issues or questions, please refer to the documentation of each toolkit or open an issue in this repository.

7. Suggested Folder Structure

Below is the recommended directory layout for this project:

ObjectDetection/
├── Python_Env/                # (Optional) Python virtual environment
├── software/
│   └── OIDv4_ToolKit/
│       └── OID/
│           ├── csv_folder/
│           └── Dataset/
│               ├── train/
│               │   └── Vehicle registration plate/
│               │       ├── images/
│               │       ├── labels/
│               │       └── Labels_Not_Yolo/
│               ├── validation/
│               └── test/
├── yolo/
│   ├── saved_model/
│   ├── sort/
│   ├── videos/
│   ├── result.mp4
│   ├── result.csv
│   ├── yolov8n.pt
│   └── *.py                  # Python scripts (e.g., main.py, util.py)