Object Detection with Rotated Bounding Boxes and Dimension Measurement

This project implements real-time object detection using YOLOv8 and OpenCV, featuring rotated bounding boxes and real-time dimension measurements. The system can detect objects, calculate their dimensions in centimeters, and provide enhanced visualization with rotation-aware bounding boxes.

Features

• Real-time object detection using YOLOv8
• Rotated bounding box detection
• Dimension measurements in centimeters
• Enhanced image processing for better edge detection
• Confidence threshold filtering
• On-screen display of measurements and confidence scores
• Real-time webcam processing

Requirements

python >= 3.8
opencv-python
numpy
ultralytics

Installation

1. Clone the repository:

git clone https://github.com/yourusername/object-detection.git
cd object-detection

2. Install required packages:

pip install opencv-python numpy ultralytics

3. Download or prepare your YOLOv8 model weights and update the model path in the code.

Project Structure

├── object_detection.py     # Main script with all implementations
├── README.md              # Project documentation
└── requirements.txt       # Project dependencies

Key Components

Image Processing Functions

• rescaleFrame(): Resizes input frames for processing
• enhance_image(): Enhances image quality using CLAHE
• get_binary_mask(): Creates binary mask for object segmentation
• get_accurate_rotated_bbox(): Calculates rotated bounding boxes

Measurement Functions

• pixel_to_cm(): Converts pixel measurements to centimeters
• calibrate_focal_length(): Calibrates camera focal length
• process_frame(): Main processing pipeline for object detection and measurement

Usage

1. Update the YOLO model path in main():

model = YOLO('path_to_your_model.pt')

2. Run the script:

python object_detection.py

3. Press 'q' to exit the program

Calibration

For accurate measurements:

1. Uncomment the calibration section in main()
2. Use a reference object of known dimensions
3. Adjust the focal_length_pixels value
4. Modify depth_cm parameter based on your setup

Best Practices

For optimal results:

• Ensure good lighting conditions
• Position camera parallel to object plane
• Maintain consistent depth for measurements
• Calibrate system with known reference objects
• Use appropriate confidence thresholds

Limitations

• Accuracy depends on camera calibration
• Fixed depth assumption (100cm default)
• Requires consistent lighting
• Performance depends on hardware capabilities

Contributing

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Create a new Pull Request

Acknowledgments

• YOLO by Ultralytics
• OpenCV community
• NumPy developers

Contact

For questions or feedback, please open an issue in the repository.