Converting an image to a Torch Tensor in Python
Last Updated :
16 Jun, 2024
In this article, we will see how to convert an image to a PyTorch Tensor. A tensor in PyTorch is like a NumPy array containing elements of the same dtypes.
A tensor may be of scalar type, one-dimensional or multi-dimensional. To convert an image to a tensor in PyTorch we use PILToTensor() and ToTensor() transforms. These transforms are provided in the torchvision.transforms package. Using these transforms we can convert a PIL image or a numpy.ndarray. The numpy.ndarray must be in [H, W, C] format, where H, W, and C are the height, width, and a number of channels of the image.
transform = transforms.Compose([transforms.PILToTensor()])
tensor = transform(img)
This transform converts a PIL image to a tensor of data type torch.uint8 in the range between 0 and 255. Here img is a PIL image.
transform = transforms.Compose([transforms.ToTensor()])
tensor = transform(img)
This transform converts any numpy.ndarray to torch tensor of data type torch.float32 in range 0 and 1. Here img is a numpy.ndarray.
Approach:
- Import the required libraries.
- Read the input image. The input image is either PIL image or a NumPy N-dimensional array.
- Define the transform to convert the image to Torch Tensor. We define a transform using transforms.Compose(). You can directly use transforms.PILToTensor() or transforms.ToTensor().
- Convert the image to tensor using the above-defined transform.
- Print the tensor values.
The below image is used as an input image in both examples:
Example 1:
In the below example, we convert a PIL image to Torch Tensor.
Python
# Import necessary libraries
import torch
from PIL import Image
import torchvision.transforms as transforms
# Read a PIL image
image = Image.open('iceland.jpg')
# Define a transform to convert PIL
# image to a Torch tensor
transform = transforms.Compose([
transforms.PILToTensor()
])
# transform = transforms.PILToTensor()
# Convert the PIL image to Torch tensor
img_tensor = transform(image)
# print the converted Torch tensor
print(img_tensor)
Output:

Notice that the data type of the output tensor is torch.uint8 and the values are in range [0,255].
Example 2:
In this example, we read an RGB image using OpenCV. The type of image read using OpenCV is numpy.ndarray. We convert it to a torch tensor using the transform ToTensor().
Python
# Import required libraries
import torch
import cv2
import torchvision.transforms as transforms
# Read the image
image = cv2.imread('iceland.jpg')
# Convert BGR image to RGB image
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# Define a transform to convert
# the image to torch tensor
transform = transforms.Compose([
transforms.ToTensor()
])
# Convert the image to Torch tensor
tensor = transform(image)
# print the converted image tensor
print(tensor)
Output:

Notice that the data type of the output tensor is torch.float32 and the values are in the range [0, 1].
Similar Reads
Python Tutorial | Learn Python Programming Language Python Tutorial â Python is one of the most popular programming languages. Itâs simple to use, packed with features and supported by a wide range of libraries and frameworks. Its clean syntax makes it beginner-friendly.Python is:A high-level language, used in web development, data science, automatio
10 min read
Python Interview Questions and Answers Python is the most used language in top companies such as Intel, IBM, NASA, Pixar, Netflix, Facebook, JP Morgan Chase, Spotify and many more because of its simplicity and powerful libraries. To crack their Online Assessment and Interview Rounds as a Python developer, we need to master important Pyth
15+ min read
Python OOPs Concepts Object Oriented Programming is a fundamental concept in Python, empowering developers to build modular, maintainable, and scalable applications. By understanding the core OOP principles (classes, objects, inheritance, encapsulation, polymorphism, and abstraction), programmers can leverage the full p
11 min read
Python Projects - Beginner to Advanced Python is one of the most popular programming languages due to its simplicity, versatility, and supportive community. Whether youâre a beginner eager to learn the basics or an experienced programmer looking to challenge your skills, there are countless Python projects to help you grow.Hereâs a list
10 min read
Python Exercise with Practice Questions and Solutions Python Exercise for Beginner: Practice makes perfect in everything, and this is especially true when learning Python. If you're a beginner, regularly practicing Python exercises will build your confidence and sharpen your skills. To help you improve, try these Python exercises with solutions to test
9 min read
Python Programs Practice with Python program examples is always a good choice to scale up your logical understanding and programming skills and this article will provide you with the best sets of Python code examples.The below Python section contains a wide collection of Python programming examples. These Python co
11 min read
Enumerate() in Python enumerate() function adds a counter to each item in a list or other iterable. It turns the iterable into something we can loop through, where each item comes with its number (starting from 0 by default). We can also turn it into a list of (number, item) pairs using list().Let's look at a simple exam
3 min read
Python Data Types Python Data types are the classification or categorization of data items. It represents the kind of value that tells what operations can be performed on a particular data. Since everything is an object in Python programming, Python data types are classes and variables are instances (objects) of thes
9 min read
Python Introduction Python was created by Guido van Rossum in 1991 and further developed by the Python Software Foundation. It was designed with focus on code readability and its syntax allows us to express concepts in fewer lines of code.Key Features of PythonPythonâs simple and readable syntax makes it beginner-frien
3 min read
Backpropagation in Neural Network Back Propagation is also known as "Backward Propagation of Errors" is a method used to train neural network . Its goal is to reduce the difference between the modelâs predicted output and the actual output by adjusting the weights and biases in the network.It works iteratively to adjust weights and
9 min read