A C++ application that processes video to measure movement of persons within the field-of-view.
This is part of a research project with Heidi Pasi, Taru Lintunen, Martin Hagger. Jyväskylä University (Finland).
More detailed information is in the CAVAPA Technical Documentation.
- Convert video to image frames using ffmpeg.
ffmpeg -i ./video.mpg -q:v 1 -qmin 1 -qmax 1 ./frames/video%06d.jpg - Run CAVAPA. It will process the images and create movement images and a CSV data file (containing the movement-measure for each frame of the video).
cavapa -i ./frames -o ./cavapa_output - [optional] Convert CAVAPA output image frames into a video using ffmpeg:
ffmpeg -i ./cavapa_output/movement%06d.jpg movement.mp4
Cavapa is a command-line executable that is written in C++ for Microsoft Windows. Cavapa uses the following software libraries:
- Multi-threaded parallel processing: Microsoft Parallel Patterns Library (PPL)
- Command-line argument parsing and file-system access: Boost C++ 1.71. boost::filesystem & boost::program_options
- Reading/writing image files (PNG, BMP, TIF, etc): Free Image & Free Image Plus.
The source code is a single file that is compiled using Microsoft Visual Studio 2019. System libraries from the Windows 10 SDK (10.0.17763.0) use the v142 platform toolset. Source code is available online at https://github.com/gregruthenbeck/CAVAPA2020 (NB: The git repo contains additional projects for charting in C# (incomplete) and processing accelerometer data (DataConverterCpp.vcxproj). The Cavapa executable is built using Cavapa.vcxproj.