Search⌘ K
Home
Courses
The Practical Guide to Python for Scientists and Engineers

Working with Scale Factor

Explore how to use the scale factor to improve face detection accuracy in images and videos. Understand the balance between detection precision and speed, and learn to tweak this parameter for better results when using Python-based machine learning algorithms in image processing.

We'll cover the following...

Face detection

In this lesson, we’ll use the image below to detect faces.

Let’s detect faces in this picture.

Python 3.8
#!/usr/bin/python
import sys
import cv2
import matplotlib.pyplot as plt
def face_detect(imgpath, nogui = False, cascasdepath = "haarcascade_frontalface_default.xml"):
    image = cv2.imread(imgpath)
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    face_cascade = cv2.CascadeClassifier(cascasdepath)
    faces = face_cascade.detectMultiScale(
        gray,
        scaleFactor = 1.1,
        minNeighbors = 5,
        minSize = (30,30)
        )
    print("The number of faces found = ", len(faces))
    for (x,y,w,h) in faces:
        cv2.rectangle(image, (x,y), (x+h, y+h), (0, 255, 0), 2)
    if nogui:
        cv2.imwrite('test_face.png', image)
        return len(faces)
    else:
        cv2.imwrite("output/Faces_found.png", image)
if __name__ == "__main__":
    face_detect(sys.argv[1])

We can see that our program detected objects that were not faces at all.

Why did this happen?

The picture seems to have been taken from afar and possibly from a mobile phone. ...