MotionEye - Student Motion Detection App

MotionEye is a PyQt6 desktop app for detecting motion in video files using frame differencing. It is a student project that runs on the CPU and does not use deep learning.

Features

• Frame differencing motion detection
• Region of Interest (ROI) selection
• Adjustable parameters:
  • Bounding box threshold (object size filter)
  • NMS threshold (suppression of overlapping boxes)
  • Motion threshold (sensitivity)
  • Gaussian blur (noise reduction)
• Save video with bounding boxes
• Optional live camera preview

Screenshots

• Main Window

• Video Loaded

• File Menu (Open Video)

• Detection In Progress (Camera)

• Parameter Bar

Quick Start

Prerequisites

• Python 3.8 or newer
• pip (Python package manager)

Installation

1. Clone or download the repository

2. Create and activate virtual environment (recommended):

   python -m venv venv

   Activate:

   • Windows: venv\Scripts\activate
   • Linux/Mac: source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

Running the Application

python main.py

Alternative command:

python -m motioneye.ui.main_window

Usage

Opening a Video File

1. Click File → Open Video (or press Ctrl+O)
2. Select a video file (supports: .mp4, .avi, .mov, .mkv)

Detection Controls

• Play/Pause: Play video with real-time motion detection
• Frame Slider: Navigate through video frames
• Start/End Frame: Set analysis range

Detection Parameters

• Bbox Thresh: Minimum detected object size (100-10000 pixels)
• NMS Thresh: Suppression intensity for overlapping boxes (0.01-0.90)
• Motion Thresh: Motion sensitivity (1-255, lower = more sensitive)
• Gaussian Blur: Smoothing filter size (1-31, odd values only)

Region of Interest (ROI)

1. Click Select ROI button
2. Draw a rectangle on the video frame to define detection area
3. Click Reset ROI to remove the selection

Saving Results

1. Click Save Video with Detections
2. Choose output location and filename
3. Wait for processing to complete

Project Structure

MotionEye/
├── main.py                      # Application entry point
├── README.md                    # This file
├── requirements.txt             # Python dependencies
├── .gitignore                   # Git ignore rules
│
├── motioneye/                   # Main package
│   ├── __init__.py
│   ├── assets/                  # Application assets
│   │   └── logo.png            # Application logo
│   │
│   ├── core/                    # Detection algorithms
│   │   ├── __init__.py
│   │   └── detection.py        # Motion detection core
│   │
│   └── ui/                      # User interface
│       ├── __init__.py
│       └── main_window.py      # Main GUI window
│
└── data/                        # Optional: sample videos
    └── samples/                 # Place sample videos here

Technical Details

Frame Differencing Algorithm

Steps used in detection:

1. Frame subtraction: absolute difference between consecutive frames
2. Gaussian blur: smoothing to reduce noise
3. Thresholding: binary mask from motion pixels
4. Contour detection: find connected regions
5. Bounding boxes: rectangles around moving objects
6. NMS: remove overlapping detections

Advantages

• Fast on CPU
• No GPU required
• No model training
• Low memory use

Limitations

• Sensitive to camera movement and lighting changes
• Works best with a stationary camera
• Parameter tuning depends on the scene

Technologies Used

• PyQt6: GUI framework
• OpenCV: Video and image processing
• NumPy: Numerical computations

Camera Support

Camera functionality is experimental and may have issues:

• Limited to live preview
• No recording support
• Less stable than video file mode
• Recommended to use video files for stable operation

Troubleshooting

Issue	Solution
No detections	Increase motion sensitivity, adjust Gaussian blur
Too many false positives	Increase motion threshold, increase bbox threshold

Contributing

Contributions are welcome. You can submit issues or pull requests.

License

This project is developed for educational purposes.

Author

Motion Detection Student Team

---

Tips: Start with default parameters and adjust for your video.