TensorFlow | How to convert image to numpy array

TensorFlow tf.keras

This article explains how to convert image to Numpy array using TensorFlow. TensorFlow provides tf.keras.utils.img_to_array method to convert image instance to a Numpy array. Below is the code snippet to convert Image to Numpy array.

Download the image with tf.keras.utils.get_file

    
import tensorflow as tf

image_file = tf.keras.utils.get_file("flower.jpg",
    "https://storage.googleapis.com/gcptutorials.com/examples/flower.jpg")
    
====== Output ======
Downloading data from https://storage.googleapis.com/gcptutorials.com/examples/flower.jpg
229376/221594 [===============================] - 0s 0us/step

Load image with tf.keras.preprocessing.image.load_img

    
img = tf.keras.preprocessing.image.load_img(image_file)

Convert image to numpy array with tf.keras.preprocessing.image.img_to_array

    
image_array = tf.keras.preprocessing.image.img_to_array(img)
print(image_array.shape)
print(image_array)
    
======= Output ======
(1297, 1920, 3)
[[[124.  73.   7.]
    [126.  75.   9.]
    [128.  77.  11.]
    ...
    [ 66.  30.  34.]
    [ 69.  30.  35.]
    [ 69.  30.  35.]]

    [[125.  74.   8.]
    [127.  76.  10.]
    [129.  78.  12.]
    ...
    [ 66.  30.  34.]
    [ 67.  31.  35.]
    [ 70.  31.  36.]]
    ...
    [[125.  76.   9.]
    [126.  77.  10.]
    [128.  77.  11.]
    ...
    [ 67.  31.  35.]
    [ 68.  32.  36.]
    [ 69.  33.  37.]]]

Complete code for converting image to numpy array

    
import tensorflow as tf

image_file = tf.keras.utils.get_file("flower.jpg",
    "https://storage.googleapis.com/gcptutorials.com/examples/flower.jpg")
img = tf.keras.preprocessing.image.load_img(image_file)

image_array = tf.keras.preprocessing.image.img_to_array(img)
print(image_array.shape)
print(image_array)