Sunglass Attach using CV2

Introduction

• In this LinkedIn article, we'll explore how to use Python's OpenCV library to automatically add stylish sunglasses to a portrait photo. Let's dive into the code and uncover the magic!
• Profile pictures are often our first impression in the digital world. Adding a twist of creativity can make your profile stand out.
• We're going to showcase how to use OpenCV, a powerful computer vision library, to programmatically place sunglasses on a person's face in an image.

Step by Step Guide

Loading Images

• The script loads a sunglass image and a destination image (the photo to which we want to add sunglasses).

sunglass_image = cv2.imread("b.png", cv2.IMREAD_UNCHANGED
    destination_image = cv2.imread("a.jpg"))        

Image

Face Detection

• The code uses a pre-trained Cascade classifier to detect faces in the destination image. If no faces are detected, the script gracefully exits.

face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
    gray_image = cv2.cvtColor(destination_image, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray_image, scaleFactor=1.1, minNeighbors=5)


    if len(faces) == 0:
        print("No face detected in the destination image.")
        return        

Positioning the Sunglasses

• Assuming a single face is detected, the script calculates the dimensions and position to place the sunglasses. The sunglasses are resized to match the face's width and positioned at a specific height above the eyes.

 sunglass_resized = cv2.resize(sunglass_image, (w, int(0.4 * h))
    x_offset = x
    y_offset = y + int(0.3 * h)        

Alpha Channel Handling

• The code checks if the sunglass image has an alpha channel (transparency). If it does, the sunglasses are blended into the destination image using the alpha channel to ensure a smooth integration. If not, the sunglasses are placed without transparency.

 if sunglass_resized.shape[2] == 4:  # Check if sunglass image has an alpha channe 1
        alpha_s = sunglass_resized[:, :, 3] / 255.0
        alpha_d = 1.0 - alpha_s
        for c in range(0, 3):
            destination_image[y_offset:y_offset + sunglass_resized.shape[0], x_offset:x_offset + sunglass_resized.shape[1], c] = (
                alpha_s * sunglass_resized[:, :, c] + alpha_d * destination_image[y_offset:y_offset + sunglass_resized.shape[0], x_offset:x_offset + sunglass_resized.shape[1], c]
            )
    else:
        for c in range(0, 3):
            destination_image[y_offset:y_offset + sunglass_resized.shape[0], x_offset:x_offset + sunglass_resized.shape[1], c] = (
                sunglass_resized[:, :, c]
            )        

Saving the Output

• Finally, the modified image with the sunglasses is saved to a file called "output_image.jpg."

cv2.imwrite('output_image.jpg', destination_image)        

Output

Feedback/Queries

I hope this article will be useful for you and that you learned something new Please, feel free to drop any questions in the comments below. I would be happy to answer them.

Thanks for reading 💚