Video Calibration Tool

Calibrate a video/image on a plane using homography.

Requirements

• Python3.x
  • opencv-python
  • numpy

Installation

If you are using conda:

conda create --name calibration python=3.8
conda activate calibration

otherwise, install your python venv (or use your system installation (r u sure tho?)).

After the env creation (both venv, conda, system, or whatever), just install the opencv bindings:

pip install opencv-python

Run

Run with:

python3 calibrate_view.py /path/to/your/video.mp4 /path/to/your/image/plane.jpg

Instructions

Press the following keys:

• "C" : calibrate the view, choose 4 points on the frame and those corresponding in the plane;
• "D" to draw up to 30 points at once to check the calibration. Press "Enter" if you want to stop before 30 points;

Re-calibrate at every run?

No! If you already did your calibration, the calibration file will be loaded, which means you can visualize more points using the "D" command.

Results

Results of the calibration will be saved in the data folder as:

data/h__cam_YOURCAMFILENAME.npy

aswell as the points used for the calibration:

data/pts1__view_YOURCAMFILENAME.npy
data/pts2__view_YOURCAMFILENAME.npy