🎥 Object tracking using optical flow

Implementation from scratch of optical flow based object tracking algorithm using C++ and OpenCV

An implementation from scratch, of an optical flow based object tracking algorithm, written in C++, for use with video files.

📘 Algorithm details

The region of interest is captured from the first, frame and is compared to each subsequent frame within a specified region around the starting point. The region is compared in grayscale to each frame to find its location.

💻 Requirements

CMake
GCC / Clang, etc.
OpenCV 4.X or newer

🚀 Usage

First, compile the program using cmake:

$   mkdir build && cd $_
$   cmake ..
$   cmake --build .

Then, run executable binary from the build directory:

$   ./build/ObjectTracker [video_file] [x_pos] [y_pos] [roi_padding] [region_padding]

🗣 Arguments

Argument	Description
video_file	The path to the video file
x_pos	The x coordinate of the centroid of the region of interest (ROI) to track
y_pos	The y coordinate of the centroid of the region of interest (ROI) to track
roi_padding	The padding around the coordinates to create square ROI
region_padding	The padding around the square ROI, to use as a search space for tracking the object

📜 Next steps