Getting Started

To compile and run the c++ implementation of CodeBooks, follow these steps:

1. Create a build directory and navigate into it:

   mkdir build && cd build
   

2. Run CMake to generate the Makefile:

   cmake ..
   

3. Compile the code:

   make
   

4. Run the program:

   ./VideoBackgroundSubtraction [-c] [-v video_path] [-o output_path] [-n new video path] [-i new background image path]
   

Command Line Arguments

• -c or -v: Choose one of these options. -c will use the camera as the video source, -v followed by the path to a sequence of frames will use that video as the source.
• -n: Use this option followed by the path to a new video if you want to change the background of a video.
• -i: Use this option followed by the path to an image file if you want to use a new background image.
• -o: Use this option followed by the path where you want to store the resulting masks. Use output_path = output/CB/{video_name} if you want to run the evaluation.

Running Python implementation of CodeBooks

python3 CodeBooks.py

• this will run CodeBooks with a video specified in the script

OpenCV built-in methods

python3 OpenCV_BS.py --source camera

or

python3 OpenCV_BS.py --source folder --path /path/to/{video_name}

• this will store the output masks in folders output/MOG/{video_name} and output/MOG2/{video_name}

Evaluation

eval_video.py

Usage:

python3 eval_video.py video_name

• will take output masks from each method (CodeBooks, MOG, MOG2) from output/{method}/{video_name} and calculate the accuracy, precision, recall and f1 scores on each of them compared to ground truth masks located in data/gt/{video_name}