Facial Landmarks and Face Detection in Python with OpenCV

Facial landmark detection is the process of detecting landmarks or regions of interest (key-points) on the face like Eyebrows, Eyes, Nose, Mouth and Jaw silhouette.

Some applications of facial landmark detection are face swap, head pose detection, detecting facial gestures, gaze direction etc.

This project was done on a jupyter notebook.

Before you start you should have the following installed on your computer:

The following packages are required, you can install them with pip from your command-line:

NOTE: The downloaded classifier and model must be moved into the same directory with your .ipynb file for this tutorial.

On your Interactive Python Notebook (*.ipynb) import the following

On line 2, convert the image from BGR to RGB colour
On line 8,9 crop image to a specific size using the image’s axes, with top-left of the image being (0,0). Use image[y:y+depth, x:x+width], where x and y are the left-hand-side axes to start the cropping. width and depth are the x and y dimensions(length) for the cropped image respectively.
Line 15 converts image to grayscale, note that face detection algorithm used here performs better on grayscale images.

Three faces are detected on the image.

Below are the detections zoomed in.

Thanks for reading!

Related posts:

Reportagem para o Jornal da PUC-Rio. “Perigo para os polegares” is published by Karen Krieger.

When people first meet me, I assume they think all kinds of things about me. I assume they think I’m too quiet or that I have an odd shirt on. I assume because I work as a mechanic people think that…

We define an employee’s total earnings to be their monthly salary* months worked, and the maximum total earnings to be the maximum total earnings for any employee in the Employee table. Write a query…