2025-07-06 Web Development
Use rembg for image background removal
By O. Wolfson
What is rembg?
rembg
is a popular Python package for automatically removing the background from images using deep learning. It’s fast, easy to use, and runs entirely on your local machine.
Key Features
-
Automatic Background Removal
rembg
detects the subject of an image and removes the background, making it fully transparent (outputs a PNG with alpha channel). -
No Cloud Uploads—Runs Locally All processing happens on your computer. No need to upload images to an external server.
-
Versatile Works with people, products, animals, objects—almost any image with a clear foreground.
-
Fast Uses pre-trained ONNX models for efficient, reliable performance (CPU or GPU).
-
Scriptable and Programmable Use from the command line or directly in your Python scripts.
How to Install rembg
To install the CLI and all extras:
shpip install 'rembg[cli]'
Note: It may be necessary to install more dependencies to get rembg to work. Read the snippet below for more details.
shpip install -r requirements.txt
How to Use rembg
1. As a CLI Tool:
Remove the background from a photo in one line:
shrembg i input.jpg output.png
input.jpg
: Your original image (JPG, PNG, or WEBP)output.png
: Output with transparent background
2. As a Python Library:
Use in your own Python scripts:
pythonfrom rembg import remove
with open("input.jpg", "rb") as i:
result = remove(i.read())
with open("output.png", "wb") as o:
o.write(result)
Typical Use Cases
- Preparing product photos for e-commerce
- Creating profile images or avatars
- Making collages or cutouts for design/art projects
- Automating bulk background removal
Why Use rembg
?
- Open Source and Actively Maintained
- Local Processing (privacy)
- Works with Python and CLI
- Great for automation and creative workflows
More Information
- Official GitHub: https://github.com/danielgatis/rembg
- Supported Formats: JPG, PNG, WEBP (input), PNG (output)
In summary:
rembg
is the easiest and most powerful way to automatically make cutouts from images, with just one command or line of code.