My roommate wants to hit one of his coworkers in the face with a Nerf gun whenever he [the coworker] walks by. This repo contains software to support that project.
You'll need to have Python 3 installed.
Once you have that, you'll need to install opencv. This can be done easily with pip:
pip install opencv-python
Currently the only method you need is facial_distance_calculator.calculate_from_image_file. You might run the following Python script in this directory:
import facial_distance_calculator
print(facial_distance_calculator.calculate_from_image_file('four_people.jpg'))
From this you would see a list of coordinates, which represents the relative location (in pixels) of the center of each face detected in the photo.
- Can we train this to look only for one individual?
- Need to run this on a raspi. That might be easy, but I've never done it. Is there configuration work to be done?
- There are some magic numbers in the face cascade matching. Are those parameters that can be tweaked/optimized?