A comprehensive Python application for real-time object detection using YOLOv8 (You Only Look Once) deep learning model. This project supports video processing, webcam detection, and batch image processing with an intuitive command-line interface.

• Video Processing: Process video files and save annotated outputs
• Real-time Webcam Detection: Live object detection from webcam feed
• Batch Image Processing: Process multiple images in a directory
• Interactive Interface: User-friendly command-line menu system
• GPU Support: Automatic CUDA detection for faster processing
• Customizable Detection: Adjustable confidence thresholds
• Visual Annotations: Bounding boxes, labels, and confidence scores
• Progress Tracking: Real-time processing progress and statistics

• Python 3.8 or higher
• CUDA-compatible GPU (optional, for faster processing)
• Webcam (for real-time detection)

1. Clone this repository:

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

2. Install the required dependencies:

pip install -r requirements.txt

3. The YOLOv8 model will be automatically downloaded on first run.

Run the main application:

python object_detection.py

1. Process Video File:

   • Input a video file path
   • Get real-time processing with live display
   • Save annotated video to output directory

2. Real-time Webcam Detection:

   • Live object detection from webcam
   • Press 'q' to quit
   • Press 's' to save screenshots
   • Option to record the session

3. Process Images:

   • Batch process images from a directory
   • Supports multiple image formats (JPG, PNG, BMP, TIFF)
   • Save annotated images to output directory

4. Exit: Close the application

object-detection/
├── object_detection.py    # Main application file
├── requirements.txt       # Python dependencies
├── README.md             # This file
├── video/                # Input videos (created automatically)
├── images/               # Input images (created automatically)
└── output/               # Processed outputs (created automatically)

The YOLOv8 model can detect 80 different object classes from the COCO dataset, including:

• People
• Vehicles (cars, trucks, buses, motorcycles, bicycles)
• Animals (cats, dogs, birds, horses, etc.)
• Household items (chairs, tables, laptops, phones, etc.)
• Sports equipment
• Food items
• And many more!

You can customize the detection parameters by modifying the VideoObjectDetector initialization:

detector = VideoObjectDetector(
    model_name="yolov8n.pt",      # Model size: n (nano), s (small), m (medium), l (large), x (extra large)
    confidence_threshold=0.25      # Confidence threshold for detections
)

• CPU: Works on any modern CPU, processing speed depends on video resolution and complexity
• GPU: Significantly faster processing with CUDA-compatible GPU
• Model Sizes:
  • yolov8n.pt: Fastest, lowest accuracy
  • yolov8s.pt: Good balance of speed and accuracy
  • yolov8m.pt: Higher accuracy, slower processing
  • yolov8l.pt & yolov8x.pt: Highest accuracy, slowest processing

1. "Could not open camera": Check if webcam is connected and not being used by another application
2. "Video file not found": Ensure the video file path is correct and the file exists
3. Slow processing: Consider using a smaller model size or reducing video resolution
4. GPU not detected: Install CUDA-compatible PyTorch version

• Use GPU acceleration when available
• Process videos at lower resolution for faster results
• Close other applications to free up system resources
• Use appropriate model size for your hardware capabilities

• Videos: Saved as detected_[original_name].mp4 in the output directory
• Images: Saved as detected_[original_name].[ext] in the output directory
• Screenshots: Saved as screenshot_[frame_number].jpg in the output directory

This project is open source and available under the MIT License.

Contributions are welcome! Please feel free to submit a Pull Request.

• YOLOv8 by Ultralytics
• OpenCV for computer vision operations
• PyTorch for deep learning framework