How to Convert Images to NumPy Array?

Images make it easier to represent the functioning model. Height, Width, and Channel formatted image data is used in Python for machine learning. Height, Width, and Channel Numpy Arrays are the new image formats. In this article, we'll examine converting images to a NumPy array.

Switching pictures over completely to NumPy exhibits is a typical undertaking in different fields, including PC vision, AI, and picture handling. It considers effective control and examination of picture information utilizing the useful assets given by the NumPy library.

To start, you'll have to import the essential libraries, for example, NumPy itself, and a picture-handling library like OpenCV, PIL, or scikit-picture. These libraries give capabilities and strategies to stack and control pictures. You can continue to stack the picture of interest when the libraries are imported. This picture can emerge out of different sources, remembering records for plate or live camera takes care of.

Contingent upon your particular prerequisites, you may likewise have to apply different activities to the NumPy cluster. Normal tasks incorporate resizing to change the picture aspects, editing to zero in on unambiguous areas of interest, or applying channels for picture upgrades or sound decreases.

At last, if vital, you can save the altered NumPy exhibit back to a picture design utilizing the capabilities given by the picture handling library, going full circle of picture control.

After loading the Image, You will transform it into a NumPy array using the image processing library's functions or methods. This change addresses the picture as a complex NumPy exhibit, where every component compares to a pixel's worth. With the picture in NumPy cluster design, you can use the broad capacities of NumPy for errands like picture examination, highlight extraction, or AI model info.

In summary, switching pictures to NumPy clusters is a basic move toward many picture-handling work processes, empowering proficient and adaptable picture examination. You can seamlessly integrate image data into your Python-based projects and easily extract valuable insights or perform various tasks by following these general steps and consulting the documentation of your chosen image processing library.

Required Modules:

NumPy: If NumPy isn't accessible (in that frame of mind of Python), one can introduce it using Python forms higher than 3. x.

Pillow: Later forms additionally require an express establishment of this. It leans toward devices for picture altering. Pillow is merely an upgrade to PIL and is a Python 3 module. It tends to set up use

Image loading through Pillow Library

How about we check whether there is a picture in the PNG or JPEG design? The picture's way may be utilized to allude to it. The center of PIL is the picture class. It includes an open () capability that opens up a picture and considers the recovery of both pixel and advanced document designs.

Image Used:

from PIL import Image
# Open the image file
image = Image.open('Sample.png')
# Get the image format
image_format = Image. format
# Get image size
image_size = Image. size
# Get image mode
image_mode = Image. mode
# Close the image file
image.close()
# Print the details
print(image_format)
print(image_size)
print(image_mode)

Output:

PNG
(400, 200)
RGB

Converting an image into a NumPy array

Python offers a wide assortment of modules and APIs to transform an image into a NumPy cluster. We should discuss Python's capacity to change over photographs to NumPy exhibits.

Converting images to a NumPy array using a NumPy module

The Numpy module itself offers various ways of achieving the same thing. These procedures are:

Example 1: Using the asarray() function

Using the asarray() function, PIL images become NumPy arrays. This capability makes an exhibit out of the info.

# Import the necessary libraries
from PIL import Image
import numpy as np
# Load the Image
img = Image.open('Sample.png')
# Convert the PIL image into a NumPy array using numpy.array()
numpydata = np.array(img)
# Check the type and shape
print(type(numpydata))
print(NumPy data.shape)

Output:

<class 'numpy.ndarray'>
(200, 400, 3)

Example 2: Using numpy.array() function

Using the function numpy. array(), which accepts an image as an argument and turns it into a NumPy array

from PIL import Image
import numpy as np
img = Image.open("Sample.png")
np_img = np.array(img)
print(np_img.shape)

Output:

(200, 400, 3)

We must print the recovered information obtained either through the asarray() method or the array() function to determine the value of each pixel in the NumPy array picture.

# Importing the necessary libraries
from PIL import Image
from numpy import asarray
# loading the Image and converting it into
# numpy array
img = Image.open('Sample.png')
numpydata = asarray(img) 
# data
print(NumPy data)

Output:

[[[111  60   0]
 [116  65   0]
 [122  69   0]
 ...
 [ 97  47   0]
 [ 99  47   0]
 [100  49   0]]
[[111  61   0]
 [118  65   0]
 [122  69   0]
 ...
 [ 97  47   0]
 [ 99  48   0]
 [100  49   0]]
[[118  65   0]
 [122  69   0]
 [126  73   3]
 ...
 [ 98  48   0]
 [100  49   0]
 [100  49   0]]
...
[[ 96  44   7]
 [ 95  43   6]
 [ 93  41   4]
 ...
 [225  80   3]
 [228  80   0]
 [229  78   0]]
[[ 93  40   6]
 [ 90  37   5]
 [ 85  32   0]
 ...
 [226  81   4]
 [231  80   1]
 [232  79   1]]
[[ 89  36   4]
 [ 84  31   0]
 [ 79  26   0]
 ...
 [228  81   4]
 [232  81   4]
 [233  80   2]]]

Getting the picture back from a Numpy Array conversion

To retrieve an image from a numpy array that has been converted, use the Image.fromarray() function. Following the back-and-forth conversion, we receive the same number of pixels. Consequently, this is quite effective.

from PIL import Image
import numpy as np
# Open the Image using Pillow
img = Image.open('Sample.png')
# Convert the Image to a NumPy array
numpydata = np.array(img)
# Check the type and shape of the NumPy array
print(type(numpydata))
print(numpydata.shape)
# Create a Pillow image from the NumPy array
pilImage = Image.fromarray(numpydata)
# Check the type, mode, and size of the Pillow image
print(type(pilImage))
print(pilImage.mode)
print(pilImage.size)

Output:

<class 'numpy.ndarray'>
(200, 400, 3)
<class 'PIL.Image.Image'>
RGB
(400, 200)

Images can be converted to NumPy arrays using the Keras API

The functions for loading, converting, and storing picture data are offered by the Keras API. Because Keras can run on top of the TensorFlow framework, having that

is essential. Images for deep learning computer vision demand Keras API. Please type the following command in the terminal to install it.

TensorFlow 2.2 or higher is required for Keras. If not, it must be installed. Please enter the following command at the terminal to install it.

pip install tensorflow

from keras. Preprocessing. image import load_img
import warnings 
# loading the Image via load_img
# function
img = load_img('sample.png') 
# Information about the picture is printed below
print(type(img))
print(img.format)
print(img.mode)
print(img.size)

Output:

<class 'PIL.PngImagePlugin.PngImageFile'>
PNG
RGB
(400, 200)

Using the Keras API, transform images to Numpy Arrays and then convert the Numpy Arrays back to images.

from keras. preprocessing.image import img_to_array
from keras. preprocessing.image import array_to_img 
# Information about the picture is printed below
print(type(img))
print(img.format)
print(img.mode)
print(img.size) 
# Convert the provided Image into a numpy array
img_numpy_array = img_to_array(img)
print("Image is converted and NumPy array information :") 
# <class 'numpy.ndarray'>
print(type(img_numpy_array)) 
# type: float32
print("type:", img_numpy_array.dtype)
# shape: (200, 400, 3)
print("shape:", img_numpy_array.shape) 
# Converting back to the Image
img_pil_from_numpy_array = array_to_img(img_numpy_array) 
# <class 'PIL.PngImagePlugin.PngImageFile'>
print("converting NumPy array into image:",
      type(img_pil_from_numpy_array))

Output:

<class 'PIL.PngImagePlugin.PngImageFile'>
PNG
RGB
(400, 200)
Image is converted, and NumPy array information :
<class 'numpy.ndarray'>
type: float32
shape: (200, 400, 3)
converting NumPy array into Image: <class 'PIL.Image.Image'>

We can verify that the source image is PIL.Image.Image and destination image types are the same as the result shown above.

Using the OpenCV Library to Convert Images to NumPy Array

The DNN and Caffe frameworks, available in OpenCV versions 3. x and later, are highly useful for solving deep learning challenges. It can be set up utilizing

The cv2 package includes the following methods.

The imread() function loads the picture and reads the supplied PIL image in NumPy array format.
The image color must then be changed from BGR to RGB.
The Image is stored in the file using imwrite().

import cv2 
image = cv2.imread('Sample.png') 
# BGR -> RGB
img = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) 
cv2.imwrite('opncv_sample.png', img)
print (type(img))

Output:

<class 'numpy.ndarray'>

Conclusion:

Python is a very flexible tool, and we have seen examples of using multiple APIs to transform photos into Numpy Arrays and back into images. By manipulating the transformed array, one can feed diverse image data into deep-learning neural networks.

Next TopicHow to Create an Animation of the Embeddings During Fine-Tuning

← prev next →