Description

This source code is for scholar practical. The goal is to compare different Kalman filter model in order to track a moving object in a video.

How to use

Build the project

Dependencies

OpenCV (Have to be installed on the system)
Boost (Installed via conan)

Instructions

Run cmake -B <path to build directory> this will create a make project and run conan install automatically
Run cmake --build <path to build directory>
Run <path to build directory>/bin/kalman_filter -h

Help text is not provided yet

Usage

First argument is the output csv file containing data about the trajectory (measured position, predicted position and corrected position). The --model-type option can be used in order to modify the kalman model type

Ex : kalman_filter data.csv --model-type=speed

To do

Help message for -h or --help.
Version message for -v or --version.
Debug & fix passing video path in arguments.
No output or output in stdout if no output path are provided.
Expose options to setup Kalman covariance matrix.
Draw on a mask instead of redrawing trajectories on each frame.