/feature-tracker

Tracking tool to extract multiple features in a video and store frame by frame ROIs in output files

Primary LanguagePython

Feature tracker

This project is a tracking tool to extract manually selected features in a video, track those features frame by frame, and store tracked ROIs in output files in the form of:

Frame_ID x1 y1 x2 y2

Required packages

This project requires Python 3.7+ to run. In order to use the tracker features of OpenCV library, install the following version of OpenCV:

$ pip3 install opencv-contrib-python==4.4.0

Usage

When running the program, two options are provided:

  • -i: path to input video file, required
  • -o: directory to store output text file, default to current directory
$ python3 main.py -i <input_video> -o <output_directory>

Output text file will be named as 'videoname_x0_y0.txt', where (x0, y0) is the top-left coordinate of selected ROI.

For example, if the following command is executed, and two ROIs are selected with (x0, y0) at (100, 200), (400, 400):

$ python3 main.py -i video/example.mp4 -o output

The following directory and files will be generated:

.
├── video
│   ├── example.mp4
└── output
    ├── example_100_200.txt
    └── example_400_400.txt

Functionality

Upon running, different mouse click events and keyboard press events have different functionalities:

Keyboard Press Functionality
SPACE Play video / pause video
ENTER Confirm selected ROI
DELETE Delete selected ROI
ESC Resume video
'q' and 'Q' Terminate program
Mouse Click Functionality
Left Button Down Select starting point of ROI
Left Button Up Select ending point of ROI
Right Button Down Select ROI for deletion