Human Body Detection Program using Python OpenCV

In the previous article, we posted about a simple face detection program using the Python OpenCV. In this article, you will learn how to detect the full human body using the Python OpenCV library.

There are a variety of computer applications that identify human bodies in digital images, like pedestrian crossing, criminal identification, healthcare, and so on. The detection program allows us to identify and locate objects. It is very important in the area of research, where the detected object can be count, accurately determined. The Python OpenCV library is mainly aimed at real-time computer vision. In this article, we are going to develop a pedestrian detection program. The OpenCV library has in-built methods to detect pedestrians.

Module Required

For the pedestrian or human body detection program, we need the following modules-

• OpenCV (cv2)
• imutils

Install OpenCV Module

OpenCV stands for Open Source Computer Vision Library. It is a free, open source library that is used for computer vision. It provides good support for the Machine Learning, Face Recognition, Deep Learning, etc.

The given command installs the OpenCV module with the pip tool. This command also installs some other supporting modules of OpenCV.

pip install opencv-contrib-python

The following code imports the OpenCV module-

import cv2 

Install imutils module

The Imutils is used for basic image processing, such as translation, rotation, resizing, skeletonization, and displaying Matplotlib images easier with OpenCV. It provides a series of convenient functions to perform operations.

pip install imutils

On successful installation, it returns the following-

Successfully built imutils
Installing collected packages: imutils
Successfully installed imutils-0.5.3

Use the following to import this library-

import imutils 

OpenCV Read and Resize Image

After module installation, we first need to read and resize the image. We will use the imread() function to read the image.

image = cv2.imread('zebracrossing.jpg') 

The loaded image can be of any size. So, we will have to resize this using the resize() method of the imutils module. If you do not want to resize the image, then you can skip this code.

image = imutils.resize(image, 
                       width=min(500, image.shape[1])) 

OpenCV HOGDescriptor

HOG (Histogram of Oriented Gradients) is an object detector used to detect objects in computer vision and image processing. The technique counts the occurrence of gradient orientation in localised portions of an image. The cv.HOGDescriptor() method creates the HOG descriptor. The hog.setSVMDetector() method sets coefficients for the linear SVM classifier.

hog = cv2.HOGDescriptor()  
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector()) 

Next, detect all the regions that have a human body inside them. The hog.detectMultiScale() method detects objects of different sizes in the input image. The detected objects are returned as a list of rectangles.

(humans, _) = hog.detectMultiScale(image,  
                                    winStride=(5, 5), 
                                    padding=(3, 3), 
                                    scale=1.21)

Next, we will draw a rectangle region surrounding each human body of the image-

for (x, y, w, h) in humans: 
    cv2.rectangle(image, (x, y),  
                  (x + w, y + h),  
                  (0, 0, 255), 2) 

Complete Code: Human Body Detection Source Code

We hope, the above code explanation help you to understand the code flow. Here, we have merged the above code chunks.

import cv2 
import imutils 
   
# Initializing the HOG person 
hog = cv2.HOGDescriptor() 
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector()) 
   
# Reading the Image 
image = cv2.imread('zebracrossing.jpg') 
   
# Resizing the Image 
image = imutils.resize(image, 
                       width=min(500, image.shape[1])) 
   
# Detecting all humans 
(humans, _) = hog.detectMultiScale(image,  
                                    winStride=(5, 5), 
                                    padding=(3, 3), 
                                    scale=1.21)
# getting no. of human detected
print('Human Detected : ', len(humans))
   
# Drawing the rectangle regions
for (x, y, w, h) in humans: 
    cv2.rectangle(image, (x, y),  
                  (x + w, y + h),  
                  (0, 0, 255), 2) 
  
# Displaying the output Image 
cv2.imshow("Image", image) 
cv2.waitKey(0) 
   
cv2.destroyAllWindows() 

The above code returns output something like this-

(env) c:\python37\Scripts\projects>pedestrain.py
Humans Detected :  3

Related Articles