Image Tuning using TensorFlow's Pixellib

By Surendhar R

Image Tuning is the Computer Vision Technique that is used to change the background of the image through Image Segmentation.

Introduction

Image Segmentation is one of the widely used applications of Computer Vision. Image Segmentation is nothing but detecting and grouping the pixels of the similar attributes of the objects in the image. It is like creating the mask for the image. In Image Segmentation, it detects the objects in the image using bounding boxes and also segments the objects in the image by grouping the pixels of the objects. There are wide applications of Image Segmentation such as cancer cells segmentation in scan reports, Image and video Tuning, etc.

Image Tuning is one of the wide applications of Image Segmentation. Image Tuning means tuning the original image and video i.e. changing the background color of the image, changing the background of the image into grayscale, changing the one background of the image using another background of the image through image segmentation.

Tech Stack and Libraries Used

Programming Language: Python.

Python Modules:

1. Pixellib
2. NumPy
3. Matplotlib

IDE Used

Google Colab

Functionality

1. Load the image
2. Pass the image to the change_bg.blur_bg() function to blur the background of the image.
3. Deeplab V3 Xception Model trained on Pascalvoc dataset is used inside the function to blur the background of the image.
4. Finally, it blurs the background of the image and gives a blurred background of the image as an output.

Results

2.

3.

4.

5.

6.

7.

8.