Face Detection Using OpenCV with Python

What are We Building?

In this blog, we will learn how to perform face detection using OpenCV and Python to detect faces in images and videos using Haar cascades. Face detection using OpenCV and Python is a popular topic in computer vision and machine learning. Haar cascades are classifiers that are trained to detect objects in images using machine learning techniques. They work by analyzing patterns in the image data and classifying the patterns as objects of interest or background noise.

Pre-Requisites

A basic understanding of computer vision and a fundamental view of what is OpenCV are needed to do this task. To know more about Computer Vision, read this article.

OpenCV is an open-source computer vision library that provides tools for image and video processing. It has a wide range of features and functions that make it popular among computer vision researchers and developers.

• Face detection

  • Face detection using OpenCV technique that involves identifying the location and size of faces in an image or video stream.
  • The most common method for face detection is to use Haar cascades, which are pre-trained classifiers that use machine learning algorithms to detect facial features such as eyes, nose, and mouth.
  • Once a face is detected, it can be extracted and processed further, for example, to recognize the person or analyze their emotional state.

• Face recognition

  • Face recognition is a more advanced computer vision technique that involves identifying and verifying the identity of a person based on their facial features.
  • Face recognition typically involves training a machine learning model on a dataset of labelled faces, so that it can learn to distinguish between different individuals.
  • The most common method for face recognition is to use deep learning models such as Convolutional Neural Networks (CNNs), which can learn highly discriminative features from face images.
  • Face recognition can be used for various tasks, such as access control, surveillance, personalization, and entertainment.

• Haar Cascades - Complete Overview

  • The basic idea behind Haar cascades is to use a set of Haar-like features to identify regions of an image that are likely to contain the object of interest. Haar-like features are simple rectangular features that can be calculated for each region of an image by subtracting the sum of pixel values in one part of the rectangle from the sum of pixel values in another part of the rectangle. These features can capture variations in brightness, contrast, and edges in the image.

  • The Haar cascades classifier is built by training a machine learning algorithm, such as AdaBoost or Support Vector Machines (SVM), on a large set of positive and negative images. Positive images contain the object of interest (such as faces), while negative images do not. During training, the algorithm learns to distinguish between positive and negative images based on the Haar-like feature and is widely used for face detection using OpenCV.

  

  • Once trained, the Haar cascades classifier can be applied to new images or video frames to detect the object of interest. The classifier works by using a sliding window over the image and evaluating the Haar-like features within each window. The resulting features are fed into the machine learning algorithm, which outputs a confidence score indicating whether the window contains the object of interest and this will be used for face detection using OpenCV.

  • One advantage of the Haar cascades approach is that it is relatively fast and can be implemented in real-time applications. However, it may not be as accurate as other methods such as deep learning approaches, particularly for complex objects with varied appearances. Additionally, training the Haar cascades classifier can be time-consuming and requires a large dataset of positive and negative images.

How Are We Going to Build This?

To perform face detection using OpenCV and Haar cascades, you can follow these general steps:

• Import the required libraries: First, you'll need to import the OpenCV library. OpenCV (Open Source Computer Vision Library) is a popular open-source computer vision and machine learning library that provides a wide range of tools and functions for image and video processing, object detection, face recognition, and more.
• Load the Haar cascades: Load the Haar cascades XML file for face detection using the cv2.CascadeClassifier() method.
• Load the image or video: Load the image or video on which you want to perform face detection using the cv2.imread() or cv2.VideoCapture() methods.
• Convert the image to grayscale: Convert the image to grayscale using the cv2.cvtColor() method.
• Detect faces using Haar cascades: Apply the Haar cascades to the grayscale image using the detectMultiScale() method of the cv2.CascadeClassifier() object. This method will return a list of rectangles representing the location and size of the detected faces.
• Draw rectangles around the detected faces: Loop through the list and draw a rectangle around each face using the cv2.rectangle() method.
• Display the image or video with detected faces: Display the original image or video with the detected faces using the cv2.imshow() and cv2.waitKey() methods.
• Clean up: Once the face detection is complete, release any resources used by the OpenCV functions and close any windows that were opened for display.

Final Output

When performing face detection using OpenCV and Haar cascades, the output typically consists of an image with rectangles drawn around the detected faces. These rectangles represent the bounding boxes of the detected faces.

It's important to note that the accuracy and quality of the face detection output can depend on a variety of factors, such as the quality of the input image, the size and orientation of the faces in the image, and the specific Haar cascade being used for detection.

Requirements

To perform face detection using OpenCV in Python using Haar cascades, you will need the following prerequisites:

• Python: First, you'll need to have Python installed on your computer. You can download the latest version of Python from the official website.

• OpenCV: You'll also need to have OpenCV installed on your computer. To perform face detection using OpenCV, we first need to install OpenCV and its dependencies. We can install it using pip by typing the following command in the terminal:

  After installing you will see an output like the image below:

• Haar cascades: Haar cascades are XML files that contain the data needed to detect specific objects in an image or video stream. For face detection, you'll need the "haarcascade_frontalface_default.xml" file. You can download this file from the OpenCV GitHub repository.

• Image or video: You'll need an image or video to perform face detection using OpenCV.

Once you have these prerequisites, you can start writing Python code to perform face detection using OpenCV (Haar cascades).

Implementation of Face Detection Using OpenCV and Python

The implementation of Face detection using OpenCV and Python is very easy and requires only a little amount of time to understand the process.

Here's a step-by-step explanation for face detection using OpenCV Haar cascades in Python:

OpenCV provides pre-trained models and APIs for detecting and recognizing objects in images and videos. This will give you access to all the functions and APIs provided by OpenCV for performing various computer vision tasks.

And, we will be using this image as input for the code that we are going to implement in the further sections.

Step 1. Import the OpenCV library for computer vision tasks.

The first step is to import the OpenCV module. This makes its functions and classes available for use in your code. Once you have imported OpenCV, you can start using its functions and classes to perform various computer vision and image processing tasks.

Step 2. Load the Haar cascades XML file for face detection using OpenCV.

The pre-trained classifiers are XML files that contain the trained models and can be loaded into your Python code using the cv2.CascadeClassifier() function.

The above code loads the XML file containing the pre-trained Haar cascade for face detection. The file should be saved in the same directory as the Python script.

The file can be viewed/downloaded from this repository.

Step 3. Load the image for face detection using opencv

To load an image for face detection using OpenCV Python, we will use the cv2.imread() function to read the image file and store it as a NumPy array.

This loads the image that you want to perform face detection on. Replace 'image.jpg' with the file name and path of your image.

Step 4. Convert the image to grayscale

To convert an image to grayscale in OpenCV Python, we will use the cv2.cvtColor() function with the cv2.COLOR_BGR2GRAY flag to convert the BGR colour image to grayscale.

This converts the loaded image to grayscale. Haar cascades are typically applied to grayscale images for faster processing mechanisms in face detection using OpenCV.

Step 5. Perform face detection using OpenCV and Haar cascades

To detect faces in an image using OpenCV Python and Haar cascades, we will use the detectMultiScale() function to perform face detection.

The scaleFactor and minNeighbors parameters control the sensitivity and accuracy of the face detection.

Step 5. Draw rectangles around the detected faces

To draw rectangles around the detected faces in OpenCV Python, loop through the list of detected faces and use the cv2.rectangle() function to draw a rectangle around each face using its bounding box coordinates.

• This loops through the list of detected faces and draws a green rectangle around each face on the original colour image.
• The (x, y) coordinates and width and height (w, h) of each detected face are obtained from the faces variable returned by the detectMultiScale() method.

Step 6. Display the image with detected faces

The final step is to display the original image with the detected faces. The first argument of the imshow() method is the name of the window that will display the image.

The output is received similarly to the image given below:

The waitKey method waits for a key press event, and the destroyAllWindows() method closes all open windows once a key is pressed.

Here's the complete code for face detection using OpenCV Haar cascades in Python:

Make sure you replace 'image.jpg' with your image's file name and path in the above code for OpenCV Python face detection.

Testing

Let’s test against another photo! The output for the above-mentioned code: